Consider moving benchmarks from tests to standalone scripts #7
Description
It makes it hard to profile the benchmarks when they are ran through the test runner - because they end up running in a separate process or isolate from the main process/isolate. It also just adds a lot of noise to stack traces and CPU profiles.
These aren't really tests anyways, they are benchmarks :).
I would suggest using a top level benchmark directory, with some benchmark scripts under there.
Sharing data with tests is the one hard thing - you can just use relative imports though (yes it's ugly, I get it, but it's fine as long as you aren't going from lib to some file outside of lib).