Temporal coherence has occasional pixel < 0

[scottstanie]

About 0.01% of pixels seem to have < 0 temporal coherence. Not sure where this is coming from yet, but could be bad pixel counting while looping over SHPs

[scottstanie]

@ehavazli has also noted that burst boundaries can have the occasional pixel equal to 1: This is likely from the same problem of having not a complete set of pixels- something like a nanmean averaging over one pixel