FEM Poisson signature changed so need to adjust the regression test of FEM Poisson timing.

[JunoRavin]

No description provided.

[manauref]

@JunoRavin
FYI to avoid future bugs
The following code

  struct gkyl_array *epsilon = mkarr1(use_gpu, basis.num_basis, localRange_ext.volume);
  gkyl_array_shiftc(epsilon, 1.0, 0);

is not equivalent to setting epsilon to 1. We need

  gkyl_array_shiftc(epsilon, 1.0*pow(sqrt(2.),basis.ndim), 0);

to take care of the normalization

[manauref]

also not ideal that no objects are released in this reg test. But maybe it's ok since it's such a small program