Adjust the resolution matrix when doing flux calibration

[segasai]

This addresses #2419

[segasai]

I think this patch is ready.
The new tests check the following.

1. If the calibration factor is constant then the resolution matrix should not change
2. If before calibration the extracted flux was subject to the condition
   $F= R * F_0$, where R is resolution matrix
   then after the calibration it should be subject to:
   $F_c = R_c * F_{0,c}$
   where $F_{0,c} = C^{-1} F_0$, and $R_c$ is the new resolution matrix

I also test the masked calibration case that it still works. For that test I use a constant calibration vector.

[coveralls]

coverage: 28.307% (-1.7%) from 30.053%
when pulling 34931c9 on fluxcalib_resol
into 38acb7f on main.