[T/THM] Use better preconditioner for PETSc

Otherwise "divergence is detected" in the first iteration
of the first time step.

Tests/Data/ThermoHydroMechanics/HeatingHomogeneousDomain/hex_THM.prj

@@ -317,7 +317,7 @@
             <petsc>
                 <parameters>
                     -ksp_type bcgs
-                    -pc_type jacobi
+                    -pc_type gamg
                     -ksp_rtol 1.e-12 -ksp_atol 1.e-14
                     -ksp_max_it 4000
                 </parameters>