Retune RICH CK theta reco. biases for DD4HEP/DetDesc individually

Small bias corrections are applied in the RICH CK photon reconstruction to the Cherenkov theta values, to account for small intrinsic biases in the reconstruction. These are small, but its useful to correct them out to maximise PID performance.

In real data, this is in effect what the n-1 scale factor run-by-run calibrations achieve. So the corrections here are really to help calibrate MC data.

Previously there was only a single set of corrections, whereas now I have slightly different values for DetDesc versus DD4HEp builds (as the underlying reasons for the tiny biases are not known its not entirely a surprise they need slightly different values).

Because this MR changes slightly the CK theta values, any QMT test running the RICH reco. will need a ref update (i.e. in Rec, Moore and Panoptes).