BUG: Assign does not take into account RestrictedFunctionSpace

[pbrubeck]

Describe the bug
Dof-wise assignment u1.assign(u2) is only well defined if u1 and u2 belong to the same FunctionSpace. However we do not check for the boundary_set of a RestrictedFunctionSpace, and here we ignore the fact that RestrictedFunctionSpace has a different DOF ordering. The assignment across different DOF orderings results in unexpected behavior.

Steps to Reproduce
Steps to reproduce the behavior:

from firedrake import *

mesh = UnitSquareMesh(3, 3)
V = FunctionSpace(mesh, "CG", 1)
Vres = RestrictedFunctionSpace(V, ["on_boundary"])

u = Function(V)
u.dat.data[:] = range(V.dim())
ures = Function(Vres)

ures.assign(u)
print("ures.assign(u)", ures.dat.data_ro)

ures.interpolate(u)
print("ures.interpolate(u)", ures.dat.data_ro)

The output is:

ures.assign(u) [ 0.  1.  2.  3.  4.  5.  6.  7.  8.  9. 10. 11. 12. 13. 14. 15.]
ures.interpolate(u) [ 3.  6.  7. 11.  0.  1.  2.  4.  5.  8.  9. 10. 12. 13. 14. 15.]

Expected behavior
Either the assignment ures.assign(u) should throw an exception, or it should apply the dof permutation and return the same output as ures.interpolate(u).

Error message
No error message.