add iterationTotal to MCMC

[danielturek]

Addresses NCT issue 591.

This PR adds an iterationTotal counter to the core MCMC run method, which tracks the total number of MCMC iterations executed (since the last call with reset = TRUE). This produces a more logical schedule of saving samples, in particular when the thin interval is not 1 and an MCMC run is continued using reset=FALSE.

See what testing turns up. I think it's all correct, but I only did very modest testing myself.

[danielturek]

@paciorek Is there a goal of getting testing to start passing again on core nimble (devel branch)?

[paciorek]

I had been planning on merging the no-laplace branch (which excises laplace) in and then moving toward release from there. I'd rather not diagnose new testing failures until laplace is excised. So I guess I will merge that in and then see where things stand.

[paciorek]

Ok, this is passing now.

[danielturek]

@paciorek @perrydv I'm ok with this getting merged. I think it's solid. That said, it does touch things at the very base.

@perrydv These changes should simplify matters for using the MCMC within the MCEM.