fix(web): show subagent models and running count in Lineage

[Bil0000]

V2 sub-agent tooltips showed "Unknown" when the sub-agent record omitted its model, even when the child thread had a model selection.

Use the child thread as a fallback and resolve model aliases with the existing composer name helpers. Both Lineage and timeline tooltips open after 200 ms instead of the default 600 ms. The Lineage title shows the running sub-agent count across all pages, includes agents awaiting a child thread, and hides the count at zero.

If neither record provides a model, the tooltip says "Not reported" instead of guessing. This changes the shared web/desktop UI; mobile has no hover tooltip.

Verified with 47 focused tests, web typecheck, and lint. Reproduced the bug and checked the fix at 1280 px and 820 px in an isolated app; measured the tooltip opening at 226 ms.

Rebased onto the current t3code/codex-turn-mapping tip after the base branch was force-pushed. The previously noted Test Server 3 failure in the unchanged queued_cancelled_while_active/codex replay scenario is resolved by ea483d1b9 ("fix(server): unblock orchestrator CI checks") on the new base.

Before:

After:

Hover timing:
Hover timing

Model: GPT-6. Harness: Codex.

[macroscopeapp]

Approvability

Verdict: Approved at f8b3940

Macroscope's review found this PR approvable — This is a small, self-contained web UI fix that improves existing subagent tooltips and Lineage status visibility without changing orchestration, schemas, product defaults, or static-analysis configuration. Focused tests cover the added display behavior.

No code changes detected at ebefc29. Prior analysis still applies.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate e323208

[Bil0000]

Rebased onto the current t3code/codex-turn-mapping tip (f180e95).

The base branch was force-pushed, which stranded this PR: GitHub compared it against a rewritten base, so it reported CONFLICTING with ~1400 changed files that were never part of this change. This PR's own commits were replanted onto the new base tip; every commit applied with no conflicts, and the diff against the new base is identical to the diff before the rebase. No behavior was changed as part of the rebase.