Consider more nuanced approach to bad pixels for acquisition selection

[taldcroft]

Currently bad pixels are handled in acq star selection by means of setting the dark current for those pixels to a very large value (700_000?) which guarantees that the pixels will not be included in a search box. This is safe, but over-conservative since some bad pixels would disrupt acquisition only if falling directly under a star.