console: configure a cluster from a drawer on the redesigned cluster page - #39992

…rm (#40043)

### Motivation

Part of the cluster details redesign stack, on top of #39992. The
configuration drawer shows a cluster's autoscaling policy, but changing
it meant copying an `ALTER CLUSTER` statement into the SQL shell. Review
feedback asked for a form. This adds the policy to the Alter cluster
form.

### Description

**The form.** Alter cluster gets an optional **Auto-Scaling Policy**
section for managed clusters on environments with autoscaling (v26.34.0
and later): an Enable checkbox, a burst size, and a linger duration such
as `15s`, `2m`, `1m 30s` or `1500ms`, empty for the environment's
`default_hydration_burst_linger`. The section is a labelled group, and
the checkbox is described by its explanation. An **Equivalent Code**
panel shows the statements a submit will run.

- The form opens on a policy it reads after opening, so a policy set or
removed moments earlier is never shown stale, from the Clusters list or
the configuration drawer. It diffs against the policy it was built from,
so a refetch landing while it's open doesn't turn an unchanged checkbox
into a change. If the policy can't be read, the section is left out
rather than shown as off.
- It names only what changes. Enabling a policy or renaming during a
graceful resize never retargets or cancels it: any SET naming `SIZE` or
`REPLICATION FACTOR` while a reconfiguration is in flight goes through
`reshape_alter_cluster_managed`, where restoring the realized values
cancels it (`cluster.rs` ~325-335). With nothing to change, the preview
says so and a submit closes the form without a request.
- The burst size defaults to the next larger size in the cluster's size
family, and to nothing at a family boundary. During an in-progress
resize it defaults to the size after the resize's target, and the target
itself is rejected while the size is unchanged, as the server does
(`HYDRATION SIZE must differ from the target SIZE ... of the in-progress
cluster resize`, `cluster.rs` ~1503). A submit that also changes Size
retargets the resize, and that SET takes the reshape path, which accepts
the old target as the burst size.
- While a resize is in flight, Size reads "Resizing to <size>." under
the field.
- From v26.35.0 a size change shows that the cluster keeps serving from
its current replicas until the new ones catch up; earlier versions keep
the downtime warning.

The modal is shared, so the classic page's Alter cluster item gets the
section and these fixes too. It used to send the realized `SIZE` and
`REPLICATION FACTOR` on every submit, which cancels an in-flight resize.
The flag behind the SQL, `enable_auto_scaling_strategy`, defaults on and
has no LaunchDarkly override.

**The SQL.** Each statement commits on its own, in order:

- **Settings:** one `ALTER CLUSTER ... SET (...)` naming only the
changed `SIZE`, `REPLICATION FACTOR` and `AUTO SCALING STRATEGY`. A
policy is `AUTO SCALING STRATEGY = (ON HYDRATION (HYDRATION SIZE =
'...'[, LINGER DURATION = '...']))`.
- **Removing a policy with other changes** joins the same SET as `AUTO
SCALING STRATEGY = ()`, so a SET the server rejects (a size that isn't
allowed, a replica or credit limit) leaves the policy in place. The
planner checks the new size against the effective config, so resizing to
the old burst size in the same statement passes. A removal on its own is
`ALTER CLUSTER ... RESET (AUTO SCALING STRATEGY)`, which also works with
the feature flag off. Neither cancels an in-flight resize.
- **Rename:** `ALTER CLUSTER ... RENAME TO ...` only for a new name; a
same-name rename writes an audit event.
- **Reads:** the form reads
`mz_internal.mz_cluster_auto_scaling_strategies` and the in-progress
target from `mz_internal.mz_cluster_reconfigurations` (`status =
'in-progress'`, `target->>'size'`) in one request. Both are fast-path
index lookups on `mz_catalog_server`, and both shipped in v26.34.0. The
request reads at strict serializable: the form makes it right after an
ALTER, and a serializable read can come back from before the write.

**Self-managed.** Self-managed has no LaunchDarkly, so a flagged UI
preview ships there outright. A preview can now set `selfManagedOptIn`,
which offers the same per-browser opt-in there: the profile menu's UI
previews and the page pill. The cluster page redesign sets it, and
`cluster-details-page-cns-173` leaves `disabledFlexibleDeploymentFlags`,
so self-managed users get the classic page until they opt in. The
cluster list preview still ships outright.

**The drawer.** The Auto-Scaling panel's copyable SET and RESET
statements become an **Edit** action that opens the form at the
auto-scaling section. The form does its own post-mount read, since it
also needs the resize target, so Edit waits only for the cluster sizes.

### Verification

- `alterCluster.test.sql.ts`, against a real Materialize:
  - setting, replacing and keeping a policy;
- a rejected removal-plus-resize leaves the policy in place, and
removing a policy while resizing to its burst size succeeds in one
statement;
- during an in-flight resize, the target is read, a burst equal to it is
rejected, and setting a policy, removing it and renaming all leave the
resize running until it finalizes at the new size;
- one SET retargets an in-flight resize and sets its old target as the
burst size;
  - settled or absent resizes read no target.
- `clusterConfiguration.test.sql.ts`: policies with and without a
linger, parsed as the drawer parses them.
- `alterCluster.test.ts`: statements for size-only, replicas-only,
strategy-only, removal with and without other changes, rename-only and
nothing changed.
- `autoScalingForm.test.ts`: linger parsing and display, size families,
the default burst size with and without an in-progress resize, and when
a submit sets, removes or keeps the policy.
- `AlterClusterModal.test.tsx`:
- the reopen sequence that used to revert a just-set or just-removed
policy now shows it and sends only the change;
  - props changing under an open form;
- no request when nothing changed, and the preview without `NaN` while
Replicas is empty;
  - the in-progress resize's target under Size;
- errors for an invalid linger, a burst equal to the cluster's size, and
one equal to an in-progress resize's target;
- the labelled group, the version-gated resize copy, no section before
v26.34.0, for unmanaged clusters, or when the policy can't be read.
- `useUiPreview.test.tsx`: the cluster page redesign is offered in
self-managed, off until the user opts in.
- `ClusterConfigurationDrawer.test.tsx`: the Auto-Scaling panel's Edit
action, and Edit waiting only for the sizes.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

<!-- devin-review-badge-begin -->

---

<a
href="https://app.devin.ai/review/materializeinc/materialize/pull/40043"
target="_blank"><picture><source media="(prefers-color-scheme: dark)"
srcset="https://static.devin.ai/assets/gh-devin-review-dark.svg?v=4"><img
src="https://static.devin.ai/assets/gh-devin-review-light.svg?v=4"
alt="Devin Review"></picture></a>
<!-- devin-review-badge-end -->

MaterializeInc#39992 typed mz_cluster_auto_scaling_strategies and
mz_cluster_reconfigurations by hand while gen:types was broken. The
generator now emits both, and AUTO_SCALING_MIN_VERSION gates them at
runtime, so the file is exactly the generator output.

### Motivation

`yarn gen:types` fails with `function "pg_get_serial_sequence" does not
exist`. kysely-codegen runs kysely's Postgres introspector for its table
query. Since the bump to kysely 0.28 (#35669), that query calls
`pg_get_serial_sequence`, which Materialize lacks. kysely 0.29 still
does. The query also lists only relkinds `r`, `v` and `p`, so the
catalog's builtin materialized views (`m`) never reached the types under
kysely-codegen 0.17. #40044's types were generated with both patched by
hand in `node_modules`.

This also bumps kysely to 0.29 and kysely-codegen to 0.20, superseding
#39686.

### Description

- `bin/kysely-codegen.ts` runs kysely-codegen programmatically with an
introspector that overrides only the table query. kysely-codegen's
materialized view query, enum handling and serializer still apply.
- The script writes the copyright header to `types/materialize.d.ts`.
The codegen output lacks it, and the copyright lint fails without it.
- `bin/generate-types` exits on errors, pulls the requested image (a
cached `latest` went stale), and stops the container on exit. Before,
each run left a container bound to port 6875, and the next run generated
types from that old container.
- kysely 0.28.17 → 0.29.8 and kysely-codegen 0.17 → 0.20.0:
- 0.29 rejects `limit(sql.raw(...))`. `rawLimit` and the lag history
limit now call `query.limit(sql.lit(n))`, which still inlines the limit
as a literal, as Materialize requires.
- `rawLimit` used to append its limit with `modifyEnd`. In 0.29,
compiling end modifiers calls `Array.prototype.toSorted`, which Chrome
109 lacks. No console query uses end modifiers now. Neither `rawLimit`
caller has an OFFSET or another end modifier, so the compiled SQL is
unchanged.
- kysely 0.29 is ESM-only and kysely-codegen 0.20 `require`s it, so
`yarn gen:types` needs Node 22.12 or later. CI and the Volta pin run
Node 26, and #39683 updates `engines.node`.
- `types/materialize.d.ts` is regenerated against v26.49.1.
kysely-codegen 0.17 and 0.20 produce identical output.
  - Adds `mz_builtin_columns` and `mz_builtin_index_columns`.
- Builtin columns lose their `Generated<>` wrappers, because #40316
changed `pg_attribute.atthasdef` to report no default for them. The
console only selects from these relations, where `Generated<T>` and `T`
read the same.
- Drops the two hand-added `TODO` markers from #39992. The generator now
emits `mz_cluster_auto_scaling_strategies` and
`mz_cluster_reconfigurations`, and `AUTO_SCALING_MIN_VERSION` still
gates them at runtime. The file is now exactly the generator output.

Worth checking on the preview, since these run the changed limits:
- the Query History list;
- a sink's Errors tab;
- the freshness lag charts on the environment overview and the cluster
pages.

🤖 Generated with [Claude Code](https://claude.com/claude-code)