Interpolation into symmetric tensor

[mscroggs]

Describe new/missing feature

Currently (after #3158), to interpolate into a symmetric tensor, we must pass in (eq) all 9 entries of a 3x3 symmetric tensor. We should additionally be able to pass in just the (eg) 6 entries of the lower part of the matrix.

See also #3143.

Suggested user interface

def test_symmetric_tensor_interpolation():
    mesh = create_unit_square(MPI.COMM_WORLD, 10, 10)

    def tensor(x):
        mat = np.array([[0], [1], [3], [1], [2], [4], [3], [4], [5]])
        return np.broadcast_to(mat, (9, x.shape[1]))

    def symm_tensor(x):
        mat = np.array([[0], [1], [2], [3], [4], [5]])
        return np.broadcast_to(mat, (6, x.shape[1]))

    element = basix.ufl.element("DG", mesh.basix_cell(), 0, shape=(3, 3))
    symm_element = basix.ufl.element("DG", mesh.basix_cell(), 0, shape=(3, 3), symmetry=True)
    space = functionspace(mesh, element)
    symm_space = functionspace(mesh, symm_element)
    f = Function(space)
    symm_f = Function(symm_space)

    f.interpolate(lambda x: tensor(x))
    symm_f.interpolate(lambda x: symm_tensor(x))

    l2_error = assemble_scalar(form((f - symm_f) ** 2 * ufl.dx))
    atol = 10 * np.finfo(default_scalar_type).resolution
    assert np.isclose(l2_error, 0.0, atol=atol)