Run some MCMC variants with no-pca and more heat flux observations

[rplzzz]

I think this may be too deep in the weeds for the paper (the last thing we need is more protocols to compare), but I'd like to see the results. What I expect to happen is that we will see a lot of (negative) correlation between the sigma values for temperature and heat flux. Basically, the MC can choose whether it wants to match temperature more accurately and sacrifice agreement on the heat flux (by making sigt small and sigf big), or vice versa. We've seen some of this kind of behavior in some of the runs we've done. It might be nice to have a concrete demonstration of it in hand.

[rplzzz]

With the demise of the protocol comparison aspect of the paper, this task is OBE.