Currently, we did not have clear idea how to record the history of fibre collision.  General speaking, we can separate information cobra-cobra and cobra-fiducial collisions.  I would suggest start from cobra-fiducial collision will be easier because fiducial fibre locations are fixed and not moving with frame. 

This could be split into two tickets: one for the cobra-fiducial fiber collisions, and one for the cobra-cobra collisions.

The cobra/cobra cobra/fid calculations are very similar.  I've pushed a draft plus notebook to tickets/INSTRM-1119 which queries the database for a given frameID, checks for cobra-cobra matches within a certain distance, and the same for cobra-fiducials. 

Note that if fibres are mid-identified (ie, when using easyID for identification), false collisions can be returned when the same spot is matched to two different cobras. This won't happen with the full identification code. 

Next step would be to add cobra-elbow collisions. 

Need to discuss whether to persist the output for this.

Will close this ticket as the code is implemented, and open another to discuss persisting hte results.