I ran some simulations with an updated Oceananigans this week and noticed that all values for the eddy viscosity coming from the dynamic Smagorinsky are zero. These simulations were working in v0.101.0.

This can actually explain #5231 and, in particular, this comment, since I remember the simulation was looking good when I changed the closure to dynamic Smagorinsky in #5124.

Taking the docs as a proxy, everything seems to be fine until at least v0.103.1.

Also, looking at the changes to the TurbulenceClosures module, the last PRs to touch it were #4892 and #5218. That as far as I looked into today. It's late here so I'll take a closer look tomorrow.

cc @glwagner @simone-silvestri @ali-ramadhan @taimoorsohail