check for OSTEP outliers

[sybenzvi]

Following up on an issue raised on slack by @LNapolitano during survey ops: on 20240909, expid 252286 a hot region was spotted in r2 during the evening darks (see below).

The OSTEP keyword in the preproc headers is supposed to track the largest step change in overscan observed in the exposure. Check for this keyword and plot the distribution of OSTEP to see if this particular exposure is an outlier. Include all exposure types and potentially sort by types.

[schlafly]

FWIW, my understanding is that OSTEP is a property of the overscan and we expect it to not depend on type of exposure. I'd hazard that the most useful dimensions are time & device (i.e., to answer questions like: "is it just r2" and "is it just recent").

[sybenzvi]

Closing this as a duplicate of desispec#2366. Will post comments there.