upgrades for cube_inject_fakedisk & pca_annular can now accept delta_sep as a list

[IainHammond]

cube_inject_fakedisk

• creation of the cube, transmission and convolution are now done in a more "accurate" order. We instead first account for transmission, then convolve with the PSF, before making the cube as recommended by the ELT experts!
• the function is significantly faster because it only convolves once
• including the transmission profile of the coronagraph is now faster and only a few lines of code, simply using np.interp as scipy's interp1d function is now legacy and not supported
• the transmission profile doesn't need to be the same size as the frame - instead, fluxes beyond the radius of the transmission curve are left untouched by multiplying by 1
• added more checks for correct inputs and updated the docstring
• added the option to not normalize the provided PSF, which in some cases might be what the user wants

pca_annular

• pretty simple one: the docstring says delta_sep can be a list, but if a list were provided there was no support for it. so now it works

[VChristiaens]

Thanks a lot @IainHammond !