Improve limiter debug info, update deformation flow example and buffers

[charleskawczynski]

This PR debugs the deformation flow example, which broke in this build (and is broken on main). So, this PR is effectively fixing CI. Closes #2151.

I've updated the way that the logging works for the limiters:

• This PR collects convergence statistics in the QuasimonotoneLimiter, and provides users with functions to print that information. This will alleviate log bloat, and give users better information for debugging.
• The convergence stats are only supported for cpu jobs, since this requires using mutable structs. Perhaps there's a way we can still do this for gpu jobs, but that will require a very different looking solution, which is not the purpose of this PR.

This test has been somewhat flaky for a while, and debugging has been challenging because the log is very bloated with false-positive warnings. The flakiness could be related to CliMA/ClimaTimeSteppers.jl#283, but I'm not sure. Ultimately, I think we need to add more precise / granular unit tests.

This PR should not be behavior changing, but I would like to, at some point, short-circuit the iteration loop in the limiters when the tracer mass (and some other variables) are zero, which results in relative error of NaNs / non-convergence warnings.