Multi-panel and multi-experiment dials.rs_mapper

[dagewa]

Allow dials.rs_mapper to work for multi-panel detectors by looping over panels at the highest level and accumulating the results. So far I tested only for a very simple 4 panel, small molecule MicroED dataset and the results appear sensible.

This should close #2333.

[dagewa]

In addition, I removed the restriction to operate over one experiment only. There may be situations like multi sweep datasets where constructing the reciprocal lattice over all sweeps is interesting.

[codecov]

Codecov Report

Merging #2362 (09e3d0b) into main (945479f) will decrease coverage by 0.02%.
The diff coverage is 52.00%.

❗ Current head 09e3d0b differs from pull request most recent head 63fa54f. Consider uploading reports for the commit 63fa54f to get more accurate results

@@            Coverage Diff             @@
##             main    #2362      +/-   ##
==========================================
- Coverage   82.83%   82.82%   -0.02%     
==========================================
  Files         593      593              
  Lines       68601    68612      +11     
  Branches     9222     9221       -1     
==========================================
  Hits        56828    56828              
- Misses       9656     9668      +12     
+ Partials     2117     2116       -1     

[biochem-fan]

Thank you very much for enhancement. The changes look OK to me.

[biochem-fan]

I think this restriction is artificial. Whether or not a multi-experiment calculation is meaningful depends on context, but we can leave that up to the user to decide.

How about adding an option to rotate the experiment according to the orientation matrix, when given
indexed.expt, not imported.expt? This might be useful when pathology (e.g. streaking) is suspected
in a serial or small-wedge crystallography dataset.

(Just an idea for future enhancement, not a requirement for this pull request.)

[dagewa]

Thanks, that sounds useful. In fact, perhaps that should be default if multiple experiments are provided that have crystal models. I might still put it in here - I can't merge this yet because we're getting failures with I23 data and it isn't clear why.