Implement new DL1 data format from ctapipe

[HealthyPear]

This will represent a major enhancement of the capabilities of protopipe and possibly other pipelines to interface with other groups and compare results between themselves.
It will also represent a major change to the code structure of protopipe (mainly write_dl1.py and the code portions called by it).

Further information in cta-observatory/ctapipe#1163.

Of course, we depend on ctapipe's own timescale.
It would be much preferable to wait for ctapipe 0.8 and import this feature straight from it (better if in sub-modules of the complete DL1 script, which currently doesn't have all protopipe's features) than hardcode it and then take it out once that release will be available.

This represents a requirement for protopipe 0.3.

[HealthyPear]

Update:

PR #58 takes already care of some stuff:

• all DL1 variables related to event data (meaning aside from metadata and such) is written by protopipe.scripts.data_training with the same names
• the HDF5 structure is still the old one, but very simple to use (and actually sufficient, since we do not work with metadata yet) - a single table

As a result, calibration (#59) and image cleaning benchmarks can now be performed using directly ctapipe-stage1-tool, in which the relevant code tested from protopipe.pipeline.event_preparer has been migrated.

ctapipe will soon have a DL1 EventSource (much like the SimtelEventSource we use to open simtel files) - see cta-observatory/ctapipe#1357.

This will be very useful to simplify things here, because:

• I can finally call ctapipe-stage1-tool as whole instead of the entire protopipe.pipeline.event_preparer and write the DL1 files (for which the only difference with the current "training" files obtained from protopipe.scripts.data_training are the table structure and the exclusive presence of DL1 information)
• I can read the DL1 file and perform the remaining part of protopipe.pipeline.event_preparer providing the shower's direction information in a separate DL2 file (so the DL1 file can be re-used for e.g. different reconstruction algorithms)
• protopipe.scripts.write_dl2 will open such DL2 file and add the remaining information (estimated particle type and energy)