TrigMuonHypo: Bug fixes in TrigMuonEFHypoTool and TrigmuCombHypoTool

I bumped into these while running some valgrind tests. I'm pretty sure the first one causes the irreproducible FPEs reported in ATR-25792 and ATR-26260. The second doesn't seem to have practical implications apart from (rightfully) upsetting valgrind. I'm not sure if anyone from the trigger side wants to take a look at this. Therefore, let me tag @tamartin.

Closes ATR-26339