Attempt to stop ODIN multiple times on unstage

We saw an issue today where during a free-run, unstaging the Eiger failed. See https://diamondlightsource.slack.com/archives/C048RD0KFFF/p1715777206781339

Broadly, our current logic is to wait for `BL03I-EA-EIGER-01:OD:NumCaptured_RBV` to be as expected then call `BL03I-EA-EIGER-01:OD:Capture.VAL 0` and `BL03I-EA-EIGER-01:OD:META:Stop 1`. We did this at 13:19:35, we then wait on `BL03I-EA-EIGER-01:OD:Capture_RBV` to go low (for 5 seconds). In this case that timed out.

@GDYendell suggests:
> The only thing I can suggest on the hyperion side is instead of a 5 second timeout, use a 1 second timeout and retry the stop 5 times. If that makes a difference it might be instructive as to what the problem is.

## Acceptance Criteria
* Implement the above
* Log when we do have to retry
* Add a task to search the logs in the middle of next run