Pk1d zchunks

[armengau]

New option in pk1d_postprocess: --apply_z_weights (default False). If set, each FFT chunck at redshift z contributes 2 nearby z bins, with a triangle-shape weight function identical to that used by the QMLE. Tests with the fujilupe mocks show the impact on the measured P1D is less than 1%.

• The zmin and zmax redshift bins must necessarily be handled differently
• This PR also includes a minor bug fix (header in output P1D FITS file).

[iprafols]

The changes look fine. I have spotted a couple of lines that potentially will need changes. @armengau can you review them? Otherwise, I think the PR is ready to merge

[armengau]

I think it's ok now