Fix stripping tests

Fixes all DaVinciTests.fsrs and DaVinciTests.stripping tests. FSR tests have been changed from running S24 to S29. This is a bit slower as S29 has a lot more lines to instantiate. The S22 test has also been changed to run S29, so is also slower. If the speed of S29 is a problem we could change these to run only a subset of the lines.

The S29 test is currently failing due to the issue with the Calo reco being run twice. The resulting FATAL errors are picked up by the test. It also has a ref file, which picks up the differences in the output, but I'm not sure we want this - the stripping output contains several things that won't be identical each time (eg, the address in memory of some algos, and the timing table). So we might want to do away with the ref file, unless these differences can be masked somehow.