Extended source spectra

[samanthalwong]

Following up on issue #42, it seems that it's not really possible to extract spectra from the ExcessMapEstimator. However, I think integral flux should be able to be extracted.

It seems that FoV background might be the best way to go about doing this (a joint spatial/spectral fit). Docs here: https://docs.gammapy.org/1.2/user-guide/makers/fov.html?highlight=fov.

[samanthalwong]

From a first look at some Crab data, the FoV spectral normalization (see fov_analysis branch) is ~10% higher than ED spectrum - might be due to inclusion of ULs but needs further validation