Update dl1 to dl2 script (Less required memory and less required step)

[SeiyaNozaki]

This PR updates the dl1 to dl2 step:

• Load each random forest model just before using it. Now the script loads all models at the beginning of the script and it consumes much memory. Just pass the model path to apply_models. This option was already implemented in apply_models, but not used for the script before.
• This script applies the particle classification models to the data twice for gammaness and reco_type. But reco_type can be set based on gammaness as suggested by @maxnoe here. Even, we can remove reco_type which will not be used for the analysis.
  Extra srcdep parameters and limit the offset of gamma MC for training #1077 (comment)

[codecov]

Codecov Report

Attention: Patch coverage is 93.75000% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 72.87%. Comparing base (1713092) to head (24ecab8).
Report is 65 commits behind head on main.

Files	Patch %	Lines
lstchain/scripts/lstchain_dl1_to_dl2.py	88.88%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1224      +/-   ##
==========================================
+ Coverage   72.67%   72.87%   +0.19%     
==========================================
  Files         132      133       +1     
  Lines       13655    13825     +170     
==========================================
+ Hits         9924    10075     +151     
- Misses       3731     3750      +19     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[maxnoe]

Nice that you get the col automtaically.. but the line above still uses the hard-coded index

[vuillaut]

@SeiyaNozaki could you move line 733 up one line and use col in 732?
After that we can merge I think

[SeiyaNozaki]

sorry I forgot to implement @maxnoe's comment.. now implemented!

[SeiyaNozaki]

Concerning the first point, it was on purpose for multiple input files...
#985

So I will modify codes like

• If there is a single input file, each model will be loaded just before it is used
• If there are multiple input files, all modes will be loaded once at the beginning and there will be no reload of models for each file

[SeiyaNozaki]

I realized I didn't read carefully comments from @gabemery during the school... sorry!

[vuillaut]

@SeiyaNozaki could you move line 733 up one line and use col in 732?
After that we can merge I think