fool-proofing statistical PSMs

[CommonClimate]

Going back to a telecon some weeks ago, I'd like to formally put on the agenda the possibility of using the "climateInterpretation" LiPD property (instance of the Interpretation class of the LiPD ontology) to help safeguard against unphysical calibrations. For instance, if the climateInterpretation_direction is negative while the slope returned by the calibration is positive, issue a warning and/or assign a zero slope to avoid the spurious calibration from pulling the analysis in the wrong direction.

I suspect this would help with our "cold MCA" problem in the northern North Atlantic, where we know from independent evidence that things were pretty toasty (else Erik the Red wouldn't have called Greenland "green").

@rtardifUW might have implemented this already; if so, please mark as resolved.