Px data

[andreufont]

Not yet a proper pull request, but we can use the tools to review the code.

Better to discuss it "in person" (Zoom), but I wanted to have a set of "PX objects" that can eventually be used in the likelihood code, but that they can also be used to do the post-processing of the raw measurements. You can do things like:

• read a PX file from @sindhusatyavolu , with original (not rebinned) measurements for each healpixel
• you can measure the mean across healpixels
• you can rebin, in theta and / or k, either the original measurements (per healpixel) or the mean
• you can compute the mean (and the covariance) of the rebinned measurements

I add a few notebooks explaining how these work. I guess the one closer to the final use would be rebinned_covariance, but the others might help you understand the intermediate steps.

I've been sloppy with the window matrix, but I believe it should be easy to finish it relatively fast. However, it would be very useful to know exactly how the matrix would be used in the likelihood code.

[andreufont]

The code at this point is more "easy to read" than efficient, it has way too many for loops. However, I don't think speed will ever be a problem in rebinning / covariance estimation, so I wouldn't worry about it. If it ever was, we could just rewrite the loops to make it a lot faster (but harder to read).