PX Pulsar distances can be negative

[blarsen10]

Some of the pulsar distances with PX measurements can jump to large negative values (albeit rarely). This appears to come about from the QuickCW.PulsarDistPriors.PXDistSampler function, since it is taking the inverse of a Gaussian distribution, which might overlap with zero if the PX uncertainty is large enough. It sounds like a possible solution to this could be just to send the prior to zero if the sampled PX value is small or negative.