Zero out parts of NIRCam grism exposures that don't have spectra

[gbrammer]

Given the long full length of dispersed spectra, some parts of NIRCam grism exposures don't have science spectra even though parts of the detector otherwise look OK, particularly for the medium filters at the red / blue edges of the LW channel.

Make a lookup table by filter/pupil/module and set some flag (e.g., DQ=1) in the exposure files to indicate that certain parts of the exposures shouldn't be used.