Slice using batch offsets fix

[jbeilstenedmands]

Missed update in #2411 as highlighted by xia2 test failures

[codecov]

Codecov Report

Merging #2472 (657c67b) into main (c706b5b) will decrease coverage by 0.01%.
The diff coverage is 100.00%.

❗ Current head 657c67b differs from pull request most recent head 81cc130. Consider uploading reports for the commit 81cc130 to get more accurate results

@@            Coverage Diff             @@
##             main    #2472      +/-   ##
==========================================
- Coverage   78.64%   78.64%   -0.01%     
==========================================
  Files         606      606              
  Lines       74245    74244       -1     
  Branches    10093    10093              
==========================================
- Hits        58393    58391       -2     
  Misses      13704    13704              
- Partials     2148     2149       +1     

[dagewa]

Thanks for finding and fixing this. I did not spot it before as I didn't run xia2 tests on the PR pair.

I have checked this and agree that your change is correct. As noted, it essentially reverts e58d8fa, one of the fixes-upon-fixes that I have been trying to disentangle in this whole process.

Imagesets are accessed using 0-based indices, and sliced thusly. The scan [ behaves in the same way, therefore calculating beg and end as 0-based and then using them for both slice operations is correct. To paraphrase the code, this is correct:

beg = sr[0] - 1 - arr_start
end = sr[1] - arr_start
exp.scan[beg:end]
exp.imageset = exp.imageset[beg:end]