PSB-85 Adds decrosstalk trigger and PipelineModule

[aamster]

The sensor that blocks the ophys_processing dag from running will be worked on separately

Also:

• refactors OphysExperiment.rois to return OphysROI objects, needed to update the decrosstalk flags. Also refactors it to pull the mask values all at once instead of 1 at a time, which is faster. Adds OphysROI.to_dict and OphysROI._mask_values
• add support generally for ophys session

Additional testing

Manually tested decrosstalk_trigger and decrosstalk (in debug mode)

[aamster]

@morriscb currently this code writes the traces into an output directory in <specimen>/<session>/DECROSSTALK/ophys_experiment_<oeid>/roi_traces.h5 for example.

This is what I think you mentioned doing in your note. I do think that decrosstalk should rather take as input the traces, rather than producing them. But that is out of scope for this PR.

What change specifically are you requesting?

[morriscb]

Nice. Sounds good, just wanted to make sure where things were going. Initially when I read over the code it didn't look like the traces were being persisted at all. Must have missed that line.

Yeah, it would definitely make sense to pull trace extraction out of this in the future and just load the previously calculated traces. I don't think there is any real difference between the methods used. Something we can do in the future as unfortunately we are hamstrung by the DECROSSTALK module being used by LIMS and any changes we make would require a change in the strategy.