Support for Digitization of Strip geometries with modules split by row

Adding functionality from !41208 (merged) in master, to allow digitization to run on geometries where we create a SiDetectorElement for each row of strips. (ATLITKSW-139)

When run on a "non split" geometry, the digitization should return compatible results as before, as can be seen here (comparing a test run using ATLAS-P2-ITK-24-00-00 with and without the changes in this MR):

http://nstyles.web.cern.ch/nstyles/dcube/digitest/

The CPU time is also not changed by this update when running on a non-split geometry (13.6 minutes to process 5k ttbar mu=0 events in both case) although there may be some CPU savings that could already be made on the current baseline performance.

Comparing a "split" geometry to a "non split" one (ATLAS-P2-ITK-24-00-00 vs ATLAS-P2-ITK-23-00-99) there are differences, but these seem to be the expected ones (e.g. changing range of eta modules, etc, for which the histograms here are not yet optimized for these updates), while things not expected to change (strip indices, barrel/endcap, etc) seem to agree well:

http://nstyles.web.cern.ch/nstyles/dcube/digitest2/

As in 21.9, a small CPU penalty is seem by running on "split" geometries (13.6 -> 14.2 minutes in a job running only ITkStrip digitization). As mentioned above though, potentially some optimizations can be made which improve both this and the baseline scenarios (e.g. data structures, geometry access, etc).

FYI @hhayward