BUG: trace space doesn't have the correct embedded_superdegree property about firedrake

[connorjward]

I fixed some of these issues in FInAT yesterday. Is your Firedrake/FInAT newer than that?
It's entirely possible that cases were missed so if this is still an issue for you I will fix it ASAP.

[tommbendall]

Yes I was hoping that our failures were all the same as #3244 -- thanks for fixing that!

I think that did fix some of our failures so I updated my FInAT last but still had these failures. I think FunctionSpace(mesh, "HDiv Trace", (h_degree, v_degree)) doesn't create a hdiv element, but actually just a plain FiniteElement

[connorjward]

OK. I'll hunt this down today 👍

[connorjward]

I have a simple FInAT fix that you are welcome to try out. This makes the initial error go away but the MFE is still not passing due to needing to add some restrictions. I'm not really familiar enough with the finite element machinery to be confident that my fix is the right thing to do though.

[tommbendall]

I have tried your branch on my local machine which fixes many of the Gusto failures -- thanks!

I think your fix is correct, I think that trace spaces on extruded meshes should have a degree that is a tuple.

I have also updated my MFE so that it is actually expected to pass. It passes on my machine with your branch (while the MFE fails when running with FInAT master and a clean cache)