Normalize edisp to integral of 1, not sum of 1

[maxnoe]

@RuneDominik does this have also implications on the interpolation code?

[codecov]

Codecov Report

Patch coverage: 99.35% and project coverage change: +0.05% 🎉

Comparison is base (19fff68) 94.73% compared to head (1164644) 94.78%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #250      +/-   ##
==========================================
+ Coverage   94.73%   94.78%   +0.05%     
==========================================
  Files          60       60              
  Lines        2981     3050      +69     
==========================================
+ Hits         2824     2891      +67     
- Misses        157      159       +2     

Files Changed	Coverage Δ
pyirf/interpolation/__init__.py	100.00% <ø> (ø)
pyirf/interpolation/utils.py	97.61% <88.88%> (-2.39%)	⬇️
pyirf/interpolation/component_estimators.py	99.27% <92.85%> (-0.73%)	⬇️
pyirf/interpolation/base_extrapolators.py	100.00% <100.00%> (ø)
pyirf/interpolation/base_interpolators.py	96.87% <100.00%> (+0.57%)	⬆️
pyirf/interpolation/moment_morph_interpolator.py	100.00% <100.00%> (ø)
pyirf/interpolation/nearest_neighbor_searcher.py	100.00% <100.00%> (ø)
...yirf/interpolation/nearest_simplex_extrapolator.py	100.00% <100.00%> (ø)
pyirf/interpolation/quantile_interpolator.py	98.24% <100.00%> (+0.09%)	⬆️
...irf/interpolation/tests/test_base_extrapolators.py	100.00% <100.00%> (ø)
... and 10 more

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

[RuneDominik]

Yes, it does. At the moment MomentMoprhInterpolator, QuantileInterpolator and MomentMorphNearestSimplexExtrapolator also norm to a sum of one. As this however only is a global factor, we could just nomalize by integration in EnergyDispersionEstimator after computing the interpolant, right?

Edit: There are probably more things that need to be changed. E.g. adding some factors to mean_std_estimation

I'm however not sure how this effects the performance of the algorithms. All EDisps I've used so far for testing summed to one as they were computed with pyirf.

[RuneDominik]

Besides some minor docstring updates and a missing test I did not find any major things here. (And ofc we need a changelog entry here, but I guess this one is obvious)

[maxnoe]

@RuneDominik done