Bugfix/noise vegas

[NickiBond]

Changes to improve how V2DL3 handles noise including:

1. Using average of qStatsData.getCameraAverageTraceVar instead of qStatsData.getCameraAverageTraceVarTimeIndpt for EA selection. This is because qStatsData.getCameraAverageTraceVarTimeIndpt excludes all pixels that were excluded at any time during the run which can lead to calculating noise from very few pixels which can be unreliable.
2. Use closest EA if noise is out of range rather than throwing out the run. A warning message telling user how far out of range it is has been added.
3. If pixels are suppressed during a timeslice, qStatsData.getCameraAverageTraceVar returns an artificially low value. Added following procedure:

If > 200 pixels suppressed:
        • Find mean and standard deviation of event noises.
        • If noise is X standard deviations lower than the mean (currently set at 3):
                • Use mean of the noises within X standard deviations instead for those noise events (usually one timeslice)
        • Else:
                • Change in noise due to suppressed pixels is small, print warning and continue.

4. Moved avNonNegativeNoises /= nNonNegativeNoises inside if/else statement to avoid a potential divide by 0 error.

[matthew-w-lundy]

This test failure is not a problem. It appears to run fine to generate files before and after, and the results are expected to be different so the comparison fails.

[NickiBond]

@matthew-w-lundy thank you for checking that.

[njucarlchen]

Code changes Look good. And conversions of a few runs on my machine succeeded without issue.