RICH - Photon pre-selection reoptimisation

A previous optimisation was slightly too aggressive in tightening the CK photon preselection cuts, leading to a signification PID degradation. This MR loosens these off a bit, returning performance to be consistent with the very loose (no cuts) configuration, as was the intention.

Nothing is for free, and this comes with a small CPU slow down. Around O(5%) on the overall RICH reconstruction sequence. However, as this almost halves the pion MisID rate for a given kaon efficiency, it is worth it, as there are better handles for reducing the CPU cost of the RICH that do not cost as much in PID performance, if required.

Refs. updated in Brunel!1085 (merged) and Moore!644 (merged).

Thanks to @bmalecki for bringing this to my attention. FYI @seaso