Sensitivity fixes

[rlopezcoto]

After debugging, together with @moralejo, the sensitivity code, I update here the code with some cleanup. Main changes are:

• Now we are using crab_hegra everywhere for the sensitivity calculations. There was some inconsistency when transforming from percentage of Crab tu flux units.
• All the weighting is done in bins of estimated energy, but the energies used for the weighting are the true energies

I also cleaned up the code, but we are still missing some variable clearer names, I'll work on it in the future. Just to report here, during all the debugging we did with @moralejo, we compared all the results at the DL2 level and beyond with those obtained using a Mars-based analysis and everything was equivalent.

[vuillaut]

Hi.

Updating the code would be the perfect moment to improve its reading with more explicit names.
I reviewed a couple of things that can be improved (probably not exhaustive).

Also, adding unit tests (especially where they were already bugs! - so they don't happen again) would improve a lot the code.

[rlopezcoto]

@vuillaut I implemented all the suggested changes, thanks very much for the review and sorry for the big delay!
I find that at the moment, a complete refactoring of this code should be done, or we can just wait for protopipe to be ready. In any case, with all the real data currently arriving, this is probably not a priority.

[vuillaut]

@vuillaut I implemented all the suggested changes, thanks very much for the review and sorry for the big delay!
I find that at the moment, a complete refactoring of this code should be done, or we can just wait for protopipe to be ready. In any case, with all the real data currently arriving, this is probably not a priority.

Hey @rlopezcoto, thank you very much.
The final goal was to have a dedicated tool for IRF production independent of other libraries (ctapipe, lstchain, etc...). In the meantime, this code will be very useful.

[rlopezcoto]

@vuillaut can we merge this one?

[vuillaut]

Hi.
there are still a few changes required for functions to work.
Could you add some unit tests, please?
-> maybe not a unit test for each function but at least a high-level one once the dl2 file has been created by the chain to make sure the functions work in principle and show how the user should make the sensitivity computation?

[SeiyaNozaki]

t is not defined now, "theta2_bins"?

[rlopezcoto]

yes

[rlopezcoto]

Thanks @SeiyaNozaki for spotting the new problems in the PR!

[rlopezcoto]

Thanks to @vuillaut for the changes as well! Not sure what is breaking the tests, but we have two options:

1. Include a dl1 proton file
2. Deactivate the test
   In any of the two cases, I would like to move on with this PR for the people to start using it and also correct subsequent errors that I'm sure will appear.

[vuillaut]

I'm looking into the unit test issue.

[rlopezcoto]

Hi, I used the creation of a fake proton file as described by @vuillaut in this PR (sorry, but I could not merge it because there were conflicts and had to solve the issues quickly).
I corrected all the issues with the units and also the problems because we were using diffuse gammas for the tests to evaluate the sensitivity. Now all checks are passing, so let's move forward with this PR.

[vuillaut]

Thank you @rlopezcoto !