Unstable forward step when PSF size is less than twice that of the image

[landmanbester]

The forward step (aka flux mop in the clean worker) is sometimes unstable when using a PSF that is less than twice the size of the image in both dimensions. I suspect this is due to the use of PCG algorithm which is very fast but not always stable. Need to investigate ways to stabilize the inversion in this case. Possible avenues to investigate:

• more stable optimisation routines (eg. the primal dual),
• shaped Tikhonov regularisation,
• BFGS-like rank 2 updates to the PSF convolution operator,
• w-binned PSFs.

The latter two, which would improve the Hessian approximation, will also make the forward step more accurate when imaging a wide field of view.

[landmanbester]

This was solved in #114. The problem was that the psf convolve operator approximating the Hessian is not Hermitian so the PCG was unstable. I believe this was also the source of the explody ripple that sometimes crept in towards later iterations. Solution is to use the absolute value of psfhat in the psf approximation. This works for reasons that are not currently very well understood (but gels well with the notion that the FT of the PSF should be approximately the gridded weights)