chore: react migration plan

)

Tasks completed: 2/2
- Port commit #50 (lore revert) and #53 (bird skill removal)
- Port commit #62 (Docker .mjs entrypoint support)

SUMMARY: .planning/phases/12-memory-qmd-hardening/12-03-SUMMARY.md

…ges to enhance branch (openclaw#50)

fix(utils): use virtual key codes for cross-layout paste

… and progressRouter (PR openclaw#51)

…openclaw#50)

Co-authored-by: Lyle <[email protected]>

…k weight

Conversation length shouldn't affect complexity scoring - message openclaw#50 isn't
inherently more complex than message openclaw#5. Users may have long conversations
with simple messages or short conversations with complex ones.

Also reduced code block weight from +10 to +5 points. Code blocks in
conversational context aren't always complex tasks - could be examples,
explanations, or simple code snippets.

This should prevent ongoing Discord conversations from artificially inflating
complexity scores and routing simple messages to expensive models.

Co-authored-by: Copilot <[email protected]>

…enclaw#50)

Reaction and block-action interaction events were enqueued via
enqueueSystemEvent but never triggered a proactive session wake.
The in-memory queue is passive — events only surface when the
next inbound message arrives for the same session. For channels
that receive triage messages with no ongoing conversation, this
means reaction/interaction system events were silently lost.

Add requestHeartbeatNow() after each enqueueSystemEvent() call
in the reaction and interaction handlers, consistent with how
exec-event, notifications-event, and other system event producers
already wake their target sessions.

Also add vi.mock for the new import in both test files to prevent
import failures.

Co-authored-by: openclaw-sre[bot] <openclaw-sre[bot]@users.noreply.github.com>