xyst

Add momentum solve for ChoCG

This changeset enables theta > 0 in ChoCG, leading to theta = 0.5
Crank-Nicholson and theta = 1.0 backward Euler. This only affects the
linear (viscous) term.

Also add back different options for advection stabilization: tg, damp2,
and damp4, yielding, respectively Taylor-Galerkin, 2nd, and 4th-order
damping. The divergence stabilization is kept at 4th-order, independent
of the stabilization of advection.

Tested with Poiseuille flow, appears correct with theta=0, 0.5, 1.0,
with damp2 and damp4 fluxes.

Tested with inviscid and viscous sphere with both quarter and full
geometry with explicit time stepping and damp4 flux, all look correct.