Add try catch to trigger control.

Also fix case where no trigger has been submitted.

This is the simplest possible attempt to work around some of the crashes we see when reading trigger control.

It seems more common now, maybe the file system's slower or maybe it's due to injections, but we're getting relatively frequent node crashes due to trigger control being read while updating.

This fixes the issue by doing a simple try catch with a few retries, then defaulting to going ahead with the upload.

Testing

I've run a test with BBH nodes and didn't encounter the crash. See /home/spiir/tdavies/runs/run-H1L1V1-BBH-tdv-try-trigger-control-1717900479-0427fb88c, though note there's a couple of restarts while waiting for the first detrspmap file to be created by a separate process.