fluids: Add diffusion to advection problem

[jrwrigh]

Adds diffusion to the advection problem (now making it a "normal" linear advection-diffusion problem). I've tested locally on an over-resolved advecting sine wave and get the analytical solution within 5 digits of accuracy over 2,000 timesteps.

[jrwrigh]

Not sure what's happening with the ROCM error. It's for a completely different fluids tests, and a BPS test failed, so it's probably unreleated.

[jeremylt]

Just for grins, do we have a cool GIF we can make with this to show off?

[jeremylt]

There's a failure on re-run, so we need to sort that out before merging

[jrwrigh]

Here's a gif for ya. IDK about cool, but it shows the diffusion dampening the higher frequency mode much faster the the lower frequency mode.

Edit: I have no idea why it looks so bad in GitHub, but fine locally. I'm guessing I missed a FFMPEG incantation to make avi -> gif not be terrible (because heaven forbid ffmpeg have decent defaults)

[jrwrigh]

There's a failure on re-run, so we need to sort that out before merging

It's non-deterministic both in whether it fails and by the error metric and what backend. Regardless, a third run through passed fine, so this should be fine now.