Fix SkyCube.spectral_index function

[OlgaVorokh]

@cdeil Hello!

Can you explain a little bit, what this function is supposed to do? Because there is no tests and usages here and you are the only one who created and modified it. Can you give me some directions in which way I should think and work on?

[cdeil]

Fermi-LAT has diffuse gamma-ray emission spectral cube models, where the data array is the flux at the given (lon, lat, energy). For every point (lon, lat) this represents a spectrum flux(energy), which is usually roughly a power-law spectrum in energy.

This SkyCube.spectral_index method is supposed to be a convenience method that let's one compute that spectral index for a given spectral cube model ... it's a quantity gamma-ray astronomers care about.

To test this, you should use this example file:
https://github.com/gammapy/gammapy-extra/blob/master/datasets/vela_region/gll_iem_v05_rev1_cutout.fits

You can try to compute the spectral index for two positions (lon, lat) at a few enerview ... the value should be similar, but not exactly the same at different positions / energies ... always roughtly the value 2.5.

And, as you already do, you should use this helper function to compute the spectral index:
http://docs.gammapy.org/en/latest/_modules/gammapy/spectrum/powerlaw.html#power_law_g_from_points

We can work on this together this afternoon if you want. :-)

[OlgaVorokh]

@cdeil, I think code is quite normal according to my understanding of spectral flux. I add a test to this method and also debug output. Maybe there is some logical problem? May you review this?

[cdeil]

I'm closing this PR now, it doesn't fix the SkyCube.spectral_index method.

We'll have to go through and clean up SkyCube some day ...