add support for different cosmics params per CCD

[sbailey]

An upcoming z5 CCD swap will use a 500 micron thick device instead of the current 250 micron devices. This leads to more diffuse cosmics, requiring different cosmics identification parameters (increasing c2fudge).

Add support to desispec.scripts.proc to use different cosmics parameters for different CCDs when doing preprocessing. This might use the CalibFinder infrastructure to add an optional override parameter (while leaving the default as something that works well for most of the CCDs).

We need this for dailyops starting in early November, but we don't need it for Himalayas/Iron which only covers earlier data.