Uncertainty convolution as a stop-gap measure during periods of few calibration response measurements

Added the ability to convolve the current uncertainty estimated from measurements with a previous suite of response curves. This is to address a particular issue with our current plan to "continuously" produce uncertainty estimates every hour.

Following major detector changes, we may only have one or two measurements to inform the uncertainty. So few measurements cannot reliably account for the true uncertainty in a detector's behavior, therefore we have decided a good ``stop-gap'' measure is to incorporate the uncertainty from a previous suite of measurements, while still updating the calibration model with the few measurements made in the new configuration. This will likely not be a fully correct uncertainty. However, at least it will provide something feasible. It is still undecided if these times will need to be re-done once more measurements are made.