Be judicious regarding the exposures which are used in the total detection image

[stscijgbot-hstdp]

Issue HLA-420 was created on JIRA by Michele De La Pena:

This issue is discussed in HLA-419 with regard to the cosmic ray rejection of the catalogs. The intent is to limit the exposures used in total detection image to avoid adding images with cosmic rays into the mix.  A snippet from HLA-419:

Recent versions of the HLA pipeline avoid creating detection images with a mix of N=1 and N>1 filters.  If all filters have N=1 exposures, then all exposures are included in the detection image.  If some filters have N>1 exposures, then only the N>1 filters get included in the detection image._  Filters in such visits that have N=1 do _not get included in the detection image.  But they still use the source lists created from the detection image for photometry on the N=1 filters.  

[stscijgbot-hstdp]

Comment by Warren Hack on JIRA:

It is not clear what should be done in the situation where filter products have regions where N=1 due to the dither pattern used. These regions will also contribute CRs to the final detection image. Also, we have to consider how the total detection image gets created, since we use 'drizzle' to create the product which will reject CRs in regions with more than 1 image (regardless of filter or dither pattern). This may vary from how the HLA generated their total detection image. Did they just add the filter products together? If so, that would explain their rejection process and why it may not be directly applicable here.

[stscijgbot-hstdp]

Comment by Warren Hack on JIRA:

Filter products where only one exposures makes up the product (N=1 products) have not had any cosmic-ray identification performed on them which serves as the underlying reason for the more complicated logic implemented by the HLA. Instead, Drizzlepac PR #⁠815 was submitted to implement single-image cosmic-ray 'rejection' for all N=1 filter products. This algorithm is an updated/expanded version of the 'classify' option used for the alignment implemented in the astrometry pipeline processing. This change updates the DQ array of the single exposures which comprise a unique filter product, which removes the identified cosmic-rays when creating the filter and total detection products.

This appears to alleviate the need for the HLA logic for handling N=1 products. Further testing should still be performed to determine whether or not this PR adequately addresses this issue.