Benchmark

- [ ] add a specific function benchmark() based on test(), with its own problem list of significant runtime. Maybe use a bench.def file which is a copy of the problem.def (or of an original version), and which can be modified before the solve, in order to broadcast options.
- [ ] make sure we can retrieve the time for the actual problem solving in the log. Check ipopt stats ? More generally, how to retrieve the benchmark results if not run locally (e.g CI run) ?
- [ ] add CI workflow when pushing on branch 'benchmark' ?