[REMI] Various fixes and improvements for quad DLDs

Description

Recent work on characterizing new (non-REMI) DLD detectors uncovered some minor annoyances in the current notebook.

• With both ignore_fel and ignore_ppl enabled, a virtual trigger is now created spanning the entire train. This is useful when data was taking using only dark hits for maximal duty cycle
• Some plots did not work for very few hits
• Chunking could fail either on axis 0 (pulses) or axis 2 (hits) when data shape is smaller than chunk shape
  • axis 0 fixed by using the maxshape=(None,) + shape[1:] trick
  • axis 2 fixed by limiting it to the detector's max_hits

How Has This Been Tested?

run = 125
in_folder = '/gpfs/exfel/exp/SQS/202331/p900382/raw'
calib_config_path = '/gpfs/exfel/exp/SQS/202331/p900382/usr/aqsx_proc_config.yaml'

karabo_id = 'SQS_AQSX_DLD4'
out_aggregator = 'DLD01'

quad_anode = Truess
ignore_fel = False
ignore_ppl = False

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

Checklist:

• My code follows the code style of this project.

Reviewers

@ahmedk @hammerd