fix(managed): tolerate an in-flight wake across the primary environment

[logancsack]

Problem

#81 opened the cached shell mid-wake, but the primary-environment auth gate still probed the guest's session state — whose transient retry budget is 15 seconds against a multi-minute wake — so a mid-wake page load surfaced "Something went wrong… fetch-session-state (HTTP 503)" instead of the workbench. Observed live on Aldo staging with the 0.0.31-devpc.16 pin.

What changed

• New isManagedWorkspaceUnavailable(): the workspace cannot serve its primary environment yet (asleep, or a wake/restart/runtime update in flight). The gateway has already authenticated the browser in every such state.
• Used wherever sleeping alone gated availability: the server-auth bootstrap bypass, the cached environment-descriptor fallback in connection registration, and the composer's waking-workspace status (which now also shows during in-flight wakes).

Verification

• apps/web full unit suite: 2088 passed (235 files), including a lifecycle-state matrix test for the new predicate.
• tsgo typecheck clean; production build succeeds.

🤖 Generated with Claude Code

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	11.8 KiB	11.8 KiB	+13 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.9 KiB	5.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.0 KiB	6.0 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.8 KiB	11.8 KiB	+6 B (+0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.9 KiB	5.9 KiB	+3 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.0 KiB	6.0 KiB	+3 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	51.4 KiB	51.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: 8749dc5 · PR result: c086cc1 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 96.9 KiB
• Claude decoded thread snapshot: 97.6 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.