Increases threshold for P1PetscApplyTest

Pipeline #16544 failed in gcc_9_eigen_petsc for
- P1PetscApplyTest1
- P1PetscApplyTest8
for the test being run on the regular_octahedron_8el.msh mesh.
Reported differences between HyTeG and PETSc result were
- 4.56077796548144763e-16 (1 MPI proc)
- 5.12848009315936904e-16 (8 MPI procs)
and, thus, above the old threshold of 4.5e-16. Looking at the
commits I cannot say where this comes from. This commit increases
the threshold to 1e-15.
23 jobs for mohr/issue99 in 84 minutes and 50 seconds (queued for 2 seconds)