fix(dispatcher): retry transient ClaimDue failures and give the claim a longer timeout

[devin-ai-integration]

Summary

On Cloud Run, cold CPU allocation and idle-pool reconnects push the first few ClaimDue calls past the 5s QueryTimeoutDuration, producing recurring failed to claim due notifications / context deadline exceeded ERROR logs even though nothing is actually wrong.

• store.ClaimTimeoutDuration = QueryTimeoutDuration*2 (10s) now bounds ClaimDue, since it opens a transaction and runs a FOR UPDATE SKIP LOCKED select plus an update rather than a single query.
• dispatchDueNotifications now claims via claimDueNotifications, which retries up to dispatcherClaimRetries (2) extra times with a dispatcherClaimRetryBackoff (2s) pause, only when isTransientClaimError(err):
  • context.DeadlineExceeded (incl. wrapped), pgconn.SafeToRetry/pgconn.Timeout, any net.Error, and PgError classes 08xx / 40001 / 40P01 / 57P01-03.
  • Anything else (e.g. 42P01, opaque errors) returns immediately, so a real outage still logs once per tick. Retries also stop as soon as the dispatcher ctx is cancelled, and a cancelled-ctx failure (shutdown) is no longer logged as an error.
• The startup sweep now runs dispatchDueNotificationsAt(ctx, startup=true): if it still ends in DeadlineExceeded after retries, it logs at Warn instead of Error.

Tests: TestClaimDueNotifications (success on 2nd attempt, wrapped deadline, retry budget exhausted, permanent PgError / opaque error not retried, no retry after ctx cancel) and TestIsTransientClaimError.

Link to Devin session: https://app.devin.ai/sessions/40339faccc2c46948d4a8fb2bee608e5
Open in Devin Desktop: https://app.devin.ai/desktop/session/40339faccc2c46948d4a8fb2bee608e5?variant=devin
Requested by: @balebbae

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring