[REMI][Correct] Propagate train timestamps to output files

Description

We noticed when getting the start time for DAMNIT that the corrected REMI data has timestamps set to 0. I'm guessing this was because this notebook predates timestamps, and it was one of the first where we actually 'write' the full output file rather than copying an input file as a starting point.

This copies across the timestamps from the input, for the relevant set of trains in each output sequence file.

How Has This Been Tested?

 xfel-calibrate remi CORRECT --cal-db-timeout 300000 --cal-db-interface 'tcp://max-exfl-cal001:8015#8044' \
  --karabo-da DIGI01 --calib-config-path /gpfs/exfel/exp/SQS/202302/p004504/usr/dld_proc_config.yaml \
  --karabo-id SQS_AQSX_DLD4 --out-aggregator AQSX01 --chunks-edges 500 7 10 \
  --chunks-amplitudes 500 7 10 --chunks-hit 500 10 --chunks-signals 500 10 --quad-anode \
  --ignore-ppl \
  --in-folder /gpfs/exfel/exp/SQS/202302/p004504/raw --run 53 \
  --out-folder /gpfs/exfel/data/scratch/kluyvert/corr-remi-p4504-r53

And checking the resulting HDF5 file.

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code follows the code style of this project.

Reviewers

@schmidtp @ahmedk @wrigleyj