PR 60326: CALSTIS should handle zero exposure time differently
PR 60327: CALSTIS sc2dcorr should use spectral order when saving the gross count rate.
Occasionally a STIS science or wavecal exposure will have zero exposure time, due for example to the take-data flag going down before the start of the exposure. In this case, calstis currently does not write the corresponding output image set (or output file, if there was only one image set in the input). This apparently causes a problem for the catalog, because there can either be a missing exposure or (if the exposure was not the last in the association) a mismatch between exposures and catalog information. Imsets with zero exposure time (except for bias images) or constant value should be written to output, but with a new header keyword IMSET_OK set to F. That is, there should be an output image set for every input image set.
A user reported that the background arrays in the x1d file for an observation of Chi Persei contained interstellar features, and those features should not have been present in the background. This was an echelle observation. When calstis does the echelle scattered-light correction (an iterative procedure), the gross count rate is determined before the first iteration, and it is saved and used for computing the background (as gross - net) after the final scattered-light iteration. Flux correction is only done after the scattered light has been removed, so all iterations are done using count rate rather than flux. In all iterations before the final calibration, 47 rows were written to a temporary x1d table. The first row was order 292, and the last was for order 338. For the final extraction, however, order 292 was skipped because that order was not present in the table (PHOTTAB) for flux calibration, and only 46 rows were written to the output x1d table. The matching of gross from the first extraction and net from the final extraction is done by row number, so the original gross and final net corresponded to different spectral orders because one row (for spectral order 292) was absent from the final table. The matching of gross from the first extraction and net from the final extraction should be done (indexed) by spectral order rather than by row number.