Improve binning and fitting of A/E histograms

[fhagemann]

Outliers at high A/E values would shift the A/E maximum-likelihood fits towards higher A/E values. Therefore, the binning of the A/E should be chosen such that these outliers are excluded when fitting Compton bands.

Here, instead of selecting the bin ranges based on quantiles, a first fit to the A/E peak is performed using curve_fit and the upper A/E bin range is set to 4σ above the main peak.

For the fitting, the likelihood is set to zero for A/E values outside of the histogram ranges.

[codecov[bot]]

Codecov Report

Attention: 15 lines in your changes are missing coverage. Please review.

Comparison is base (dd1cf0e) 0.39% compared to head (0e16267) 0.38%. Report is 1 commits behind head on dev.

Files	Patch %	Lines
src/aoefit.jl	0.00%	13 Missing :warning:
src/aoe_calibration.jl	0.00%	2 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## dev #28 +/- ## ======================================== - Coverage 0.39% 0.38% -0.01% ======================================== Files 19 19 Lines 1024 1031 +7 ======================================== Hits 4 4 - Misses 1020 1027 +7 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[oschulz]

@theHenks ?

[theHenks]

Let's merge that PR and then all the rest of the Vancouver CM presentation for a new release. @oschulz is that ok?

[oschulz]

Sure!

[oschulz]

You need to resolve the merge conflicts first, though. :-)

[fhagemann]

These changes are already on the dev branch of legend-exp/LegendSpecFits.jl, so merging dev into main should be the way to go