Remove the top level --boot flag from ResumeActorRequest - #1548

Resume resolves its restore source in strict priority order: the
actor's node-local snapshot, then its own durable external snapshot,
then the template's golden snapshot, then a cold boot from the
template spec. The boot flag was consulted at exactly one point in
that chain, where it suppressed the golden-snapshot fallback, which
made its behavior much narrower than "boot from scratch" suggests:

- Actor has its own external snapshot, boot=true: flag ignored,
  restores the actor's snapshot.
- Actor has a local snapshot, boot=true: flag ignored, restores the
  local snapshot.
- Actor has no snapshot, template has a golden snapshot: boot=false
  restores the golden, boot=true cold boots from the spec. The only
  case the flag ever changed.
- Actor has no snapshot, template has no golden snapshot: cold boot
  from the spec regardless of the flag.

The one behavior the flag still controlled is not worth a public API
field: when a template has no golden snapshot, the golden actor is
cold booted from the template spec anyway, so the cold-boot path
remains exercised without a caller-facing escape hatch. The flag was
silently ignored in every other state, no caller outside the glutton
benchmark ever set it, and no test covered boot=true.

The glutton benchmark was the only caller that set boot=true, on each
actor's first resume, to report true cold-start latency as a separate
ResumeActorColdStart stats row. That row is renamed to
ResumeActorFirstResume: a fresh actor's first resume now restores the
template's golden snapshot instead of cold booting, so the old name
would misrepresent what it measures.

The proto field number and name are reserved per convention.