chore(quest): retire flate-binary on main and track the JS closed-track leak

[kixelated]

Problem

Cleanups from this session's merge reports:

• feat!: fold moq-binary into moq-flate #4425 folded moq-binary into moq-flate on dev, but main still lists quest/m1/flate-binary.md as open.
• Merge main into dev #4428 found that on dev, a subscriber joining an already-closed JS track with unlimited retention is never removed from the cache.

Changes

• Delete quest/m1/flate-binary.md and its m1 README bullet. The references now point at feat!: fold moq-binary into moq-flate #4425 on dev, since main's code still has moq-binary. m2/flate/bindings.md loses the blocker.
• New m1 [XS] quest js-closed-track-leak.md (targets dev).

Decisions

Small cleanups

• ✅ Delete flate-binary on main
• ✅ test-flakes-2 bullet for a_relay_on_the_line_answers_broadcast (already on main via test(tokio): match a worker group's full address when counting its sockets #4404; nothing to add)
• ✅ js/net closed-track leak quest
• Wildcard line wording (not selected)

Public API / wire impact: none (quest files only).

(Written by Claude Opus 5.5)

🤖 Generated with Claude Code

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-29T12:13:00.086641Z	342bf25	PR opened

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[kixelated]

MERGE

Positive improvement: yes. #4425 already folded moq-binary into moq-flate on dev, so keeping quest/m1/flate-binary.md open on main is stale bookkeeping. Retiring that quest, rewiring the m1/m2 references to #4425, dropping the m2 flate bindings blocker, and filing the XS js-closed-track-leak follow-up from the #4428 merge report are the right cleanups.

Worth the complexity: yes. Quest markdown only (25/61), no public API or wire impact. The new quest is scoped to dev (where unlimited retention lives), names the late-subscriber path in js/net/src/track.ts, and asks for a cache-empty test once the subscriber drops.

Different approach: none needed. Leaving flate-binary open until main gets the fold would keep a done item on the m1 board and force every cross-link to point at a dead file. Capturing the closed-track leak as its own XS quest is better than burying it in the sync PR notes.

Ship it.

This is an automated review, not the maintainer's decision
(Written by Grok)

[kixelated]

Merge summary:

• No changes needed. Codex found nothing (👍 on 342bf25); the Grok review had no findings.
• Checked references: feat!: fold moq-binary into moq-flate #4425 (moq-binary into moq-flate) and Merge main into dev #4428 (main into dev) are both merged into dev, and no flate-binary.md references remain.
• The new quest's leak is real on dev: #addSink in js/net/src/track.ts registers a late sink's mirrors in the cache but only attaches the cleanup hook while the track is open.
• quest check passes; CI is green.
• The decisions in the body stand as recorded.

Enabling squash auto-merge.

(Written by Claude Opus 5.5)