Tolerate slow or unreachable cross-node SharedDoc during session cleanup

[midigofrank]

Description

The collaborative editor Session calls SharedDoc.unobserve/1 during cleanup, which issues a cross-node GenServer.call to a SharedDoc that may live on another node. When that node is unreachable (:noconnection) or slow to reply (:timeout), the call exited and crashed the Session; a Phoenix channel blocked in save_workflow then inherited the exit and dropped the client's editor connection.

Wrap both unobserve call sites (terminate/2 and the parent :DOWN handler) in safe_unobserve/1, which catches the exit and logs at warning instead of crashing. Skipping the unobserve is safe: the SharedDoc monitors each observer and runs the same cleanup from its own :DOWN handler when the Session dies.

Closes #4817

AI Usage

Please disclose whether you've used AI anywhere in this PR (it's cool, we just
want to know!):

• I have used Claude Code
• I have used another model
• I have not used AI

You can read more details in our
Responsible AI Policy

Pre-submission checklist

• I have performed an AI review of my code (we recommend using /review
  with Claude Code)
• I have implemented and tested all related authorization policies.
  (e.g., :owner, :admin, :editor, :viewer)
• I have updated the changelog.
• I have ticked a box in "AI usage" in this PR

[github-actions]

The PR changes are scoped to a Session cleanup resilience fix — safe_unobserve/1 catches :exit from cross-node SharedDoc.unobserve/1 in terminate/2 and the :DOWN handler. No data access, web entrypoint, or persistence paths are touched.

Security Review ✅

• S0 (project scoping): N/A — change only wraps a cleanup-path GenServer.call in lib/lightning/collaboration/session.ex:140,438 with an :exit rescue; no new queries or entrypoints touch project-scoped data.
• S1 (authorization): N/A — no new web-layer actions, handle_events, or controllers; the modified code is internal GenServer cleanup invoked by process lifecycle, not user requests.
• S2 (audit trail): N/A — no inserts/updates/deletes on configuration resources; the patch only changes a failed cleanup unobserve into a logged no-op (lib/lightning/collaboration/session.ex:156).

[midigofrank]

@stuartc , I'm merging this in. There is a follow up issue to handle the cause of the crash #4877 . Also, the failing lint check isn't related to this PR: #4878