OpenMP: add respective pragmas to remaining loops over primitives, and possibly more
With !372 (merged) we introduced OpenMP support.
In general, everything should be working fine, but there is potential for much more parallelization.
Foremost, pragmas have been added to the loops over the macro-primitives, which is probably the easiest, safest way to implement shared-memory parallelism to the framework.
This issue shall maintain an (extensible) list of functions that could be very well parallelized, but where the pragmas are still known to be missing.
- remaining loops in vertexdof-, and edgedof-function
- some 2D operators
- most/all non-constant operators