Consolidate the project roadmap and promotion gates for core, Topology and optional telemetry #45

The 23 October review covers the seven org repositories, seven related forks, all open issues/PRs, and fresh read-only checks of live/dev/dev1/Canadaverse. Use this as the coordination point for the project roadmap in docs #43, with optional collection contracts in #44.

The main problem is promotion and acceptance, not a shortage of feature branches: most open Atlas/Topology/backend PRs overlap. The core candidates are server #266/#267 and web #182/#183; the optional collection candidates are server #268/#269, web #184/#185 and Flutter #11. Accept one sequence and retire superseded drafts only after the accepted commits preserve their required behavior.

• Establish a usable dev1 acceptance environment. Its four containers are stopped and the public site returns 502; a branch named dev1 is not a deployed candidate.
• Qualify the core database/runtime transition and performance budgets. Live still has three migration records and four large-cursor failures; dev has schema 053. Keep the cursor-only hotfix option independent of slower feature qualification.
• Separate shared-host capacity from query/render cost. Preview I/O pressure is high, global route reads time out, and observer-directory traffic ranking returned a retryable-load failure as HTTP 500. Do not call a noisy Pi result a universal software benchmark.
• Complete build-tool dependency remediation and clean CI. Web #186 currently fails after passing assertions because a deferred virtualizer callback runs after jsdom teardown. The current core lockfile has development dependency advisories; the production-only audit reports none.
• Complete basic Atlas/Topology UI and performance acceptance, preserving route ambiguity, sample labels, camera state and live effects. Adopt the bounded Docker log settings prepared for docs Bound Docker container logs in all deployment templates #13 without changing live containers in this review.
• Define sender/event/receipt time and delayed MQTT replay policy for server #231/#270, including durable acceptance and retention/deduplication boundaries.
• Qualify optional telemetry separately: physical BLE/Wi-Fi/background tests, intake lifecycle/operating ownership and shared RF coordination before broad rollout. The app replaces a polling client, not the intake service.
• Publish the exact paired release manifest, migration/recovery runbook and operator sign-off before live promotion. Keep production changes separate from this review.

The handoff will contain a disposition for every current issue and PR, repository/environment identities, owners by role, acceptance gates and the proposed merge order. No reviewer ping, org merge, production write or automatic closure is implied by this coordination issue.