Compression test

[mcclearyj]

This branch adds infrastructure necessary to produce MEDS files similar to those expected during flight. Changes include:

• Simulated exposures now include randomly-sampled dark frames
• medsmaker now includes a quick-reduce function that can accommodate median dark subtraction and a flat-field correction (right now there's none but the attribute exists).
• The medsmaker_mocks.py makes a combined BPM+master_dark mask, similar to the one planned for OBA doc (ref)
• It also saves a combined weight for single-band coaddition (like we discussed here)
• The output SWarp WCS for single-band coadds has been changed to TAN from TPV. Since PIFF knows how to handle TAN (but not TPV), PIFF can be now run on the coadd and its PSF included in the MEDS if use_coadd=True. This should address Figure out what placeholder to put in the coadd PSF image #83
• There is a boolean flag to control whether raw or calibrated images are saved to the MEDS file (hardcoded for now but doesn't have to be)

Known issues:

• Because the meds package multiplies image cutouts by a floating-point value before saving a MEDS, int16-format images cannot be saved. Doing so will require modification of a few lines in the meds/maker.py code here.
• The combined bad pixel mask/dark image is currently used as a weight for single-epoch image source extraction. SExtractor cannot handle weight images in the "0==good, 1==bad" format, so the combined mask is saved in "1==good, 0==bad" format. However, meds expects masks with the "0==good, 1==bad" convention. As a workaround, a mask is saved in each format.