feat(js_run_devserver): notify after runfiles sync

[longlho]

Why

js_run_devserver copies runfiles into a sandbox. iBazel's notify_changes_v1 reports workspace changes before that copy finishes, so dev servers cannot safely use those paths for HMR.

What

Adds opt-in notify_runfiles_changes support:

• requests ibazel_notify_changes_v1
• selectively syncs affected source runfiles while still scanning generated outputs
• falls back to a full sync for graph or unmappable changes
• emits JS_RUN_DEVSERVER_SYNCED after sync, with added, changed, and deleted sandbox paths plus the triggering iBazel event
• sets JS_RUN_DEVSERVER_NOTIFY_RUNFILES_CHANGES=1 for the child process

Existing targets keep legacy behavior unless they opt in.

Test

bazel test //js/private/test/js_run_devserver:all //js/private/devserver:watch_checked_tests --test_output=errors

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3f1b569eb6

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Fall back when iBazel omits the v1 event

When notify_runfiles_changes is enabled with an older iBazel version that recognizes ibazel_notify_changes but not ibazel_notify_changes_v1, this branch buffers every legacy success marker and returns, while syncBuild is invoked only if a structured event later arrives. No such event will arrive in that environment, so incremental outputs remain stale and the child never receives IBAZEL_BUILD_COMPLETED SUCCESS; preserve legacy synchronization, fail fast on an unsupported protocol, or add a bounded fallback instead of waiting indefinitely.

Useful? React with 👍 / 👎.