fix: prevent timer leak in GTFS retry loop

[ashnaaseth2325-oss]

SUMMARY

This PR fixes a timer lifecycle issue in InitGTFSManager where retry backoff used time.After, which cannot be stopped on context cancellation. It replaces those calls with time.NewTimer to ensure proper cleanup.

The changes affect internal/gtfs/gtfs_manager.go, specifically the retry sleep logic in the GTFS initialization flow.

---

FIX

// Before
select {
case <-ctx.Done():
    return nil, ctx.Err()
case <-time.After(delay):
}

// After
retryTimer := time.NewTimer(delay)
select {
case <-ctx.Done():
    retryTimer.Stop()
    return nil, ctx.Err()
case <-retryTimer.C:
}

---

VERIFICATION

Test by starting the service with a failing GTFS endpoint so it enters the retry loop, then cancel the context (e.g., send SIGTERM) while it is sleeping. The function should exit immediately with ctx.Err() instead of waiting for the delay. Retry behavior should remain unchanged when the context is not cancelled, with no lingering timers.

[ashnaaseth2325-oss]

Hello @aaronbrethorst
Kindly review this PR whenever you have time.

[ashnaaseth2325-oss]

Hi @aaronbrethorst
Just a gentle ping whenever you have time