extraction of timestamps should be from volume_frames

[samupicard]

for single depth mpci data, neural_frames and volume_frames are equivalent. But for multi-depth data, they are not (e.g. for 2 planes, there are twice as many edges in neural_frames than in volume_frames).

Because the time extractor works on neural_frames only, we have issues with multi-plane data, e.g. the mpci.times dataset ends up being twice as long as the mpci.ROIActivityDeconvolved, mpci.ROIActivityF and mpci.ROINeuropilActivity datasets.

I believe a simple fix would be to change this line to read out volume_frames instead of neural_frames here, but there may be other places where this needs work.