Radius of excluded regions might need to be energy dependent

[HealthyPear]

I guess this can be a second-order improvement with respect to more pressing issues.

The angular extension of some sources is not stable with energy so e.g. in general at higher energies a fixed exclusion (circle) region radius might cause an excessive amount of data discarded from the background estimation.