Fix ell0 in PS from file and broadcast amplitude

[dpole]

I think that the following eval function for PowerSpectrumFromFile has two bugs

fgspectra/fgspectra/power.py

Line 84 in 9d9fe8e

return amp * self._cl[..., ell] / self._cl[..., ell_0, np.newaxis]

1. It adds a new axis to the evaluation at ell = ell_0 for unclear reasons
2. It doesn't rise errors when ell_0 == None, the unintended result is the addition of an extra axis to the template Cl (in addition to the axis added at the previous point)

I propose to remove ell_0 and use the amp argument

[dpole]

@xzackli , these changes possibly affect some of your case-of-use. Can you review them and, in case you agree, fix the ACT tests accordingly?

[dpole]

I think this PR should be closed. @xzackli , as I said in my last comment, I believe it would be better if you could give it a look before we merge: some of your tests/applications may be affected