Add guard to GetTime

[rpitelka]

When a waveform analyzer does not reconstruct any PEs for a given DigitPMT, any event fitters that call GetTime from the FitterInputHandler will cause a crash because GetTime expects a non-empty vector of times.

To fix this, I added a check to GetTime in FitterInputHandler that dies when the function is called when no PEs were reconstructed. I also added checks to all the fitters that call GetTime to skip PMTs with no reconstructed PEs.

I'm not sure this is the best solution, so input is appreciated, but it does fix the known crash when empty waveform analysis results from RAVEN (caused by the max_total_charge parameter being set too low) are fed into quad.

[JamesJieranShen]

Discussed offline -- I'm not sure if throwing out all the hits above max charge is the best option. I think it may be better to have InputHandler fall back to digit* when the fit returns empty, needs more thinking....

[JamesJieranShen]

Thanks @rpitelka , are the changes to FitDirectionCenter now obsolete?

[rpitelka]

@JamesJieranShen Fixed revert, now only FitterInputHandler is modified by this PR

[syjeon162]

Looks good! Can we add some documentation about this behavior wherever relevant? (Maybe the FIT_COMMON or DIGITIZER_ANALYSIS ratdbs?)