Use bin edges internally for frequency-dependent specific energy

[astrofrog]

This implements the suggestion in #267 to use bin edges instead of centers as the canonical way to represent the bins to use for the frequency-dependent specific energy.

[astrofrog]

@dnarayanan - just FYI. Is the slight API change an issue for you? (you'd need to use centers= in set_specific_energy_spectrum_bins). Or would you actually prefer to pass bin edges?

[dnarayanan]

@dnarayanan - just FYI. Is the slight API change an issue for you? (you'd need to use centers= in set_specific_energy_spectrum_bins). Or would you actually prefer to pass bin edges?

it would be fine for me!

[astrofrog]

Just to double check, do you rely on the behavior of the end bins being catch-all in the sense they extend to +/- infinity? Or would it also work for you to specify only edges and have those always be finite? (Just to know if we can simplify some code)

[dnarayanan]

finite would actually work just fine!

[astrofrog]

@dnarayanan - ok thanks! So just to confirm, after the cleanup here which does simplify a bunch of things, the API is e.g.:

m.conf.output.output_specific_energy_spectrum = 'last'  # (or 'all')
m.set_specific_energy_spectrum_bins(np.logspace(11., 16., 101))

where the second method is given the bin edges (so n+1 values for n bins). Can you confirm this is ok for you? (before I merge this and make a new release of Hyperion soon)

[dnarayanan]

yep confirmed - this will be fine for me thank you!