This pull request adds automatic performance benchmarking to the tests to prevent performance degradation.
Using pytest-benchmark, it collects performance data on three real-world benchmark search spaces.
It also runs a microbenchmark to get a system performance factor.
If python-constraint performance on these benchmarks is lower than expected based on the reference results relative to the system performance factor, it throws an error.
Benchmark results are processed by the GH action benchmark and reported in the test workflow.