events: fix goroutine leak with until flag in file backend

[HindzStark]

- What I did
Fixed a goroutine and memory leak in the file events backend when using the --until flag.

Previously, libpod/events/logfile.go spawned an unmanaged goroutine using time.Sleep(time.Until(untilTime)) that ignored context.Context cancellation. If a client disconnected early (e.g. dropped API connection or Ctrl+C), the goroutine was orphaned and continued sleeping in the background.

I replaced the blocking time.Sleep with a time.NewTimer and a select block that listens for ctx.Done(). This ensures the goroutine exits immediately if the client cancels the request, properly freeing system resources and preventing an API denial-of-service vector.

- How I did it

• Replaced time.Sleep with time.NewTimer.
• Added a select statement listening to <-ctx.Done() and <-timer.C.

- How to verify it

1. Run podman system service
2. Fetch events via the API with a far-future until parameter: curl -v --unix-socket /run/user/1000/podman/podman.sock "http://d/v4.0.0/libpod/events?until=9999h"
3. Terminate the curl request (Ctrl+C).
4. Verify via pprof or logging that the background routine is destroyed and no longer sleeping.

- Description for the changelog

Fixed a goroutine leak in the event logger that occurred when an event stream using the `--until` flag was prematurely cancelled by the client.

**- Fixes** 
Fixes #29491

[Honny1]

We have been getting a lot of low quality PRs that do not follow our contribution guidelines and LLM policy. Because our time is limit I reserve the right to just close these PRs without further comments.