Caching image_response.contents in data interface for deconvolution

[augustus-thomas]

Description of Change

The data interface class for DC2 (DataIF_COSI_DC2 inheriting from ImageDeconvolutionDataInterfaceBase) now has attribute image_response_contents initialized to image_response.contents. In this way we no longer retrieve image_response.contents every iteration in the following deconvolution methods:

• calc_T_product (MStep)
• calc_expectation (post_processing)

Significance

These changes result in a 10x speedup in image_deconvolution. Instead of retrieving image_response.contents before every tensordot, we now only retrieve it once during initialization.

Profiling "no save" (before this change) versus "save" (after this change)

Other Considerations

This could also be solved by setting track_overflow=False on the image_response matrix using histpy==1.1.4

Change does not increase RAM usage. We had previously already allocated memory to the contents during deconvolution.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 73.32%. Comparing base (d02846a) to head (bf79422).
Report is 36 commits behind head on develop.

Files with missing lines	Coverage Δ
cosipy/image_deconvolution/dataIF_COSI_DC2.py	90.90% <100.00%> (+0.06%)	⬆️