BUG: Horizontal R space on tensor product element is not globally constant #5394

Describe the bug
This is a bug only on the horizontal R space - not the vertical.

I would like to construct an element on a cuboid mesh that is a tensor product of R space and CG.

elt = TensorProductElement(FiniteElement("R", quadrilateral, 0),
FiniteElement("CG", interval, p))
M = FunctionSpace(mesh, elt)

This TensorProductElement treats the R space as a per-cell discontinuous constant.

Steps to Reproduce
The following code compares the actual number of DOFS for this TensorProductElement to the expected DOFs (P*Layers + 1 when CG^P is in the extrusion direction and R in the horizontal direction).

We only extrude in 2 layers for each example so the number of DOFs coming from CG should not change. If the R space works correctly then increasing the resolution of the base mesh should not change the number of DOFs.

from firedrake import *
P = 2
LAYERS = 2
EXPECTED = P * LAYERS + 1
print(f"{'nx,ny':>8} {'base cells':>11} {'R (x) CG':>10} {'DQ0 (x) CG':>11} {'expected':>9}")

for n in (1, 2, 3, 4):
    base = UnitSquareMesh(n, n, quadrilateral=True)
    mesh = ExtrudedMesh(base, layers=LAYERS, layer_height=1.0 / LAYERS)

    cg_v = FiniteElement("CG", interval, P)
    r_elt = TensorProductElement(FiniteElement("R", quadrilateral, 0), cg_v)
    dq_elt = TensorProductElement(FiniteElement("DQ", quadrilateral, 0), cg_v)

    dim_r = FunctionSpace(mesh, r_elt).dim()
    dim_dq = FunctionSpace(mesh, dq_elt).dim()

    print(f"{f'{n},{n}':>8} {n*n:>11} {dim_r:>10} {dim_dq:>11} {EXPECTED:>9}")

Expected behavior
I expect every example R \times CG example in the above to have 5 DOFs.

Environment:

• OS: MACOS
• Python version: 3.14.6