feat(scheduler): every background run is a durable row

What a job last did lived in the API process's memory and was gone at
every restart; the audit log kept a run only when it was manual, failed
or changed something. Now each run is a job_runs row written at its
start and its end in its own session: who started it, its outcome as
data, what it spent of the YouTube quota (quota_events.job_run_id) and
the lines it logged (log_events.run_id, filterable in the log console).

- a run a stopped process left open is marked interrupted at startup
- "run now" is a one-shot job on the scheduler, which now always starts
  (interval jobs only with SCHEDULER_ENABLED); the manual actor is reset
  after the run, since pool threads are reused
- the dashboard's last run comes from job_runs, so it survives a deploy
- runs leave the audit log: migration 0109 moves the old job_run rows
  over (and back on downgrade); the audit page's run folding goes
- job_run_retention_days (default 30) prunes old runs in audit_gc

backend/tests/test_admin_logs.py

@@ -24,6 +24,7 @@ def _row(**kw) -> dict:
         "request_id": None,
         "user_id": None,
         "job_id": None,
+        "run_id": None,
         "version": "0.90.0",
     }
     row.update(kw)
@@ -80,6 +81,7 @@ def test_newest_first_and_every_row_once_across_pages_even_on_equal_timestamps(d
         ({"event": "none"}, {"warn"}),
         ({"request_id": "req-abcdef12"}, {"mine"}),
         ({"job_id": 42}, {"err"}),
+        ({"run_id": 7}, {"warn"}),  # a background run's lines, not a download job's
         ({"q": "FORBIDDEN"}, {"err"}),  # in attrs, any case
         ({"q": "KeyError"}, {"warn"}),  # in the stack
         ({"q": "100%"}, {"plain"}),  # a literal, not a wildcard
@@ -102,7 +104,14 @@ def test_each_filter_selects_exactly_its_rows(db, admin, make_user, params, expe
             job_id=42,
             attrs={"code": "forbidden"},
         ),
-        _row(msg="warn", level="WARNING", sev=13, event=None, exc="Traceback …\nKeyError: 'x'"),
+        _row(
+            msg="warn",
+            level="WARNING",
+            sev=13,
+            event=None,
+            exc="Traceback …\nKeyError: 'x'",
+            run_id=7,
+        ),
         _row(msg="plain", ts=T0 - timedelta(hours=2), attrs={"note": "100% done"}),
         # Matched by no filter above but `until`: another account, and `100%` read as a pattern.
         _row(msg="decoy", ts=T0 - timedelta(hours=3), attrs={"note": "1000 done"}, user_id=99),