RichDet improvements for new PMT classes
This MR contains a number of bug fixes and improvements, driven by the work started by @bmalecki and and extend by myself, to support processing data with the new RICH PMT number scheme.
A number of bugs have been found, some of which affect the old PMT data. In particular it was found that an inconsistency in the way the PD smartIDs where set in a number of places, had a strong impact on the likelihood background algorithm. This MR (and a related one in REC to follow soon) addresses this.
There are still a few issues to be addressed in RichDet, for instance the PD IDs saved in the DB need to have their 'large' PMT falgs properly set, and likewise the readoutChannelList methods need to include this flag. However, I would prefer to get the changes so far, as they are numerous, committed, and work on these in smaller follow up MRs.
One nice improvement, for both old and new PMTs is I realised findPMTArraySetupSIMD could be speeded up as in most cases the pixel level information is not needed, such as in PDWindowPointSIMD, and thus skipping computing this gives a nice speed bost to the RICH ray-tracing as used by the reconstruction.
@cattanem This and the REC MR will cause diffs in the tests, probably reasonably significant.