Consider both normalized and unnormalized filtering criteria when evaluating trace peaks

[PMSeitzer]

Update SECTrace class and logic to consider normalized and unnormalized intensity thresholds.

The only difference between the normalized and unnormalized rawIntensities is the division by a big constant, which should probably become a field, e.g. totalIntensitySum.

All of the smoothing, peak-picking, and spectral matching logic need not change - just needs to be restructured to consider this constant when making parameter assessments.

Finally, if the params->traceNormalizeToSumIntensity flag is true, This needs to not affect the actual rawIntensities vector, but rather, downstream outputs. Previously, this had been done at an early stage, but this doesn't need to be true.

There doesn't need to be a SECTrace::normalizedRawIntensity or SECTrace::normalizedSmoothedIntensity vector, since the only difference between them is the totalIntensitySum constant.

This needs to be done carefully, considering every place the intensity is referenced. For example, SECTrace::getFragment() needs to be updated.