Fit Px from uncontaminated mocks that used weights from contaminated mocks

[andreufont]

In order to understand the difference between Px measured from contaminated and uncontaminated mocks, @sindhusatyavolu has recently re-run the measurement on uncontaminated mocks when recycling the weights from the contaminated mocks.

It would be important to fit both and compare their best-fit models.

[andreufont]

The measurement are in
/global/cfs/cdirs/desi/users/sindhu_s/Lya_Px_measurements/mocks/analysis-49/
and we need to compare
uncont_w_contweights/bf3_binned_out_px-zbins_4-thetabins_20_w_res.hdf5
with
uncontaminated/bf3_binned_out_px-zbins_4-thetabins_20_w_res.hdf5

[andreufont]

You can find a new notebook to investigate this in notebooks/analysis/investigate_mocks.ipynb.

The results are suspicious as expected. The best-fit value of bias is very different when using the two different sets of weights.

I think we have to debug the Px measurement code, or the code to compute the window matrix.