BUG: LinearSolver always uses a non-zero initial guess

[jrmaddison]

Describe the bug
The ksp_initial_guess_nonzero parameter is ignored by LinearSolver.

Steps to Reproduce

from firedrake import *

mesh = UnitIntervalMesh(10)
space = FunctionSpace(mesh, "CG", 1)
test, trial = TestFunction(space), TrialFunction(space)

solver = LinearSolver(assemble(inner(trial, test) * dx),
                      solver_parameters={"ksp_type": "preonly",
                                         "pc_type": "jacobi",
                                         "ksp_max_it": 1,
                                         "ksp_initial_guess_nonzero": False})
b = assemble(inner(Constant(1), test) * dx)

u = Function(space, name="u")

u.assign(0)
solver.solve(u, b)
print(u.dat.data_ro)

u.assign(1)
solver.solve(u, b)
print(u.dat.data_ro)

displays

[1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5]
[1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1.]

Expected behavior
The non-zero initial value should be ignored in the second case, the linear solver should use a zero initial guess, and the results of the two solves should be the same.