Correct order of writing in add_two_photon_series

[h-mayorquin]

As discussed with @CodyCBakerPhD it seems that the write_imaging pipeline and in particular the add_two_photon_series function is writing our images inverted.

This PR is connected to the proper correction inroiextractors and the requirements consequently pinned to them:
catalystneuro/roiextractors#136

[h-mayorquin]

The changes introduce on this were already merged as of #546. We should use this PR to pin the requirements to a specific commit once we merge catalystneuro/roiextractors#136 in roieextractors .

[CodyCBakerPhD]

Leaving this open until we get a chance to talk about it with Ben...

Be prepared for the possible answer 'NWB ophys (x,y) values actually mean (cols,rows)' lol. We really should just use better language on the NWB schema, so thanks for getting that conversation started.

[h-mayorquin]

Leaving this open until we get a chance to talk about it with Ben...

Be prepared for the possible answer 'NWB ophys (x,y) values actually mean (cols,rows)' lol. We really should just use better language on the NWB schema, so thanks for getting that conversation started.

NeurodataWithoutBorders/nwb-schema#522

[CodyCBakerPhD]

NeurodataWithoutBorders/nwb-schema#522

Yep, that's what I meant. Thanks for linking.

TBH my own opinion is that on the schema side, it might make the most sense to a new user to refer to their underlying image as 'width' x 'height', since those are quantities they can think about without referencing the data storage (to them it may only appear as an picture on their screen, they may never even interact with it programmatically).

[CodyCBakerPhD]

Turns out NWB expects width x height, which is the roiextractors equivalent of cols x rows, hence the transposition here.