23.0-bugfix-PixelConditionsData

Addresses ATLASSIM-6660 and the comments made in !64185 (merged). In the (rare) case that charge calibration resulted in a large charge result from the legacy fit (in excess of 1e5), and a linear fit was resorted to but failed due to the divisor in the calculation being zero, the existing logic in rel.23 returned the legacy fit result anyway. In my refactoring, I had inadvertently changed that to return the 'invalid' value of 0.0. Logic is here introduced which reverts to the previous behaviour.