No ispyb entry on failed pin tip centre

[DominicOram]

We saw a bug in https://ispyb.diamond.ac.uk/samples/sid/5054617 where the sample failed to centre and produced no ispyb entry saying what the issue was.

The logs suggest that Hyperion thought that the pin tip centre was correct but then failed to find the tip again for the grid calculations. By looking at the images for the rotation scan in ispyb (which is where Hyperion moved to after pin tip centre) it seems that the pin tip centring was actually quite far off (I suspect due to #924).

Acceptance Criteria

• In this case we still get an ispyb entry with an appropriate error
• We make the pin tip centre more robust (maybe as a hotfix?) Done in Pin tip centring now takes a median of values dodal#242

[DominicOram]

For hotfix of pin tip centre suggestion is to:

• Take a handful of readings then take the median in x and y to find a close centre.
• Log the points and the deviations
• Take images for each and put them somewhere on /dls/

[DominicOram]

Ispyb error currently doesn't happen as we do the grid detection before we set up the xray centre grid callback. I think we should rethink this after we have the ispyb deposition on a separate process @dperl-dls

[DominicOram]

Decided against saving images as until we do #938 we're not really guaranteed the image is correct anyway

[DominicOram]

Once we have robot load in Hyperion the error should go in that