Cleanup of Gammapy-data

[AtreyeeS]

This lists some proposals for cleanup of gammapy data. There are some confusing/unreadable files that might be better to remove.

1. estimators/pks2155_hess_lc/pks2155_hess_lc.fits updated to have (at least) 2 bins in energy
2. Background files in tests/irf/ lack energy units. leads to UnitConversionError. There is in fact a TODO in /gammapy/makers/tests/test_map.py:520 to create a better example file.
3. Remove model.xml in tests/models/light_curve
4. Remove xml files in tests/models. Also remove the gc fits files in same location. These files are not used, and relevant files already exist
5. Inconsistency between column & header names in 1es0229_hess_spectrum tests/spectrum/flux_points. The README says these files were downloaded from the GADF, but the referenced GADF link does not exist any more. Remove files.
6. The stat_scan values in tests/spectrum/flux_points/binlike.fits are not compatible with the results. Remove? This is used in the docstring for FluxPoints.recompute_ul but can be replaced by the examples in estimators
7. The data available in tests/unbundled/tev_spectra can be useful. The are presently unused, but it may be useful to document this somewhere (in gammapy.astro maybe). Or, we can remove it...

This is separate from the issue of the catalogs we can provide (ie, all fermi catalogs or only the latest one, etc)

If this sounds okay, I can do the cleanup. Opinions @registerrier @QRemy @bkhelifi @MRegeard @Astro-Kirsty @adonath

[Astro-Kirsty]

Thanks @AtreyeeS. The list looks good to me!