test_level2_image_processing_pipeline and test_level2_grism_processing_pipeline lead to unexpected "okify" results

[braingram]

The test_level2_image_processing_pipeline and test_level2_grism_processing_pipeline tests are very long, containing many asserts and more than 1 call to compare_asdf.

This is an issue for "okifying" results. For example, if test_level2_image_processing_pipeline fails on the first compare_asdf:

romancal/romancal/regtest/test_wfi_pipeline.py

Line 46 in a8d1007

assert diff.identical, diff.report()

and the result is "oked". The next regression tests will likely continue to fail because the second compare_asdf:

romancal/romancal/regtest/test_wfi_pipeline.py

Line 258 in a8d1007

assert diff.identical, diff.report()

is never reached and the second file is never produced for "okifying".

It would be beneficial to break up these long tests to avoid needing to okify several regression test runs (and to make the test output more informative).