Benchmark various parallel options
We have numerous levels of parallelism that interact:
1. The MATLAB PCT
2. `set_dynare_threads`
3. parallel MATLAB instances spawned by `meson` in the testsuite
4. BLAS-level parallelism
We need to systematically investigate whether our default settings make sense and are efficient. That requires:
- [ ] Creating a parallel test group in the testsuite (see https://git.dynare.org/Dynare/dynare/-/work_items/1970#note_25217 lists for included routines)
- [ ] systematically benchmarking the settings on different machines
issue