Record image transform for every frame in a recording.

[campagnola]

The primary use is for tracking the focus depth for every frame in a stack. If the transform changes rapidly, then there will be some lag between each frame and its recorded transform (but we can live with this).

Complications:

1. Potentially very expensive
   • Do we record the entire transform stack for each frame?
   • Can we optionally just record the global transform, or just the global translation, or just the focal depth?
2. Messes with MetaArray (currently can't record more than one axis value per frame)
3. Messes with ImageView (want to display depth OR time on the z axis)

Suggestions:

1. Just record the global translation (x,y,z) values
2. Add a new key to the metaarray info: metaarray._info[0]['translation'] = array((N,3)). When we record image stacks, we currently append a new frame time to _info[0]['values'] whenever a new frame comes in. We can extend this mechanism to optionally append values to the translation field as well. Perhaps it would look something like ma.write(filename, appendAxis='Time', appendKeys=['translation'])) (this would be in acq4/util/imaging/record_thread.py : writeFrames() )
3. We can kick this one down the road. Maybe it doesn't need to be handled by ImageView at all..