fix: keep alert id stable when labeling a solo sequence FP

[MateoLostanlen]

• When a sequence labeled non-wildfire is the only one in its alert, the previous flow detached it, deleted the (now empty) alert, and created an identical replacement — churning the alert id for no reason.
• Now the solo case is detected via a LIMIT 1 siblings probe and the alert is left untouched.
• Multi-sequence alerts still split out the FP into its own alert so it stops contributing to triangulation.
• Adds an integration test for the stable-id case and a unit test for the early-return branch; updates the existing multi-sequence unit test to mock the new probe.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.11%. Comparing base (b4fa415) to head (c6e93b9).
⚠️ Report is 8 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #584      +/-   ##
==========================================
+ Coverage   89.80%   90.11%   +0.30%     
==========================================
  Files          55       55              
  Lines        2422     2508      +86     
==========================================
+ Hits         2175     2260      +85     
- Misses        247      248       +1     

Flag	Coverage Δ
backend	?
client	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fe51]

Thanks a lot for the PR @MateoLostanlen ! tested and work on devenv

Just a comment and suggestion, not blocking but wondering of would more relevant. Actually, current isolation is probably fine but worth tightening since the fix is trivial. Tell me what do you think

[fe51]

The two assertions query the entire Alert and AlertSequence tables, so they implicitly assume no other rows exist in the DB at that point. If test isolation ever weakens (shared fixtures, parallel runs, leftover data), these will fail for the wrong reason.

Scoping both queries to the sequence under test makes them resilient to whatever else is in the DB:

Fetches only AlertSequence rows for THIS sequence, checks it still points to the original alert

Suggested change

	alerts_res = await detection_session.exec(select(Alert).execution_options(populate_existing=True))
	alerts_rows = alerts_res.all()
	assert [row.id for row in alerts_rows] == [original_alert_id]
	alert_links_res = await detection_session.exec(
	select(AlertSequence.alert_id).where(AlertSequence.sequence_id == seq.id)...
	)
	assert set(alert_links_res.all()) == {original_alert_id}

[MateoLostanlen]

Thanks @fe51 I added your change

[fe51]

All good to merge now !