A commandline flag to override pytrace in failure

[FortStatement]

I use pytest to automatically check my students' exercises.
They receive the output from pytest.fail, and I don't want them seeing the entire stack trace, including local paths, potential edge cases they're supposed to find themselves, et cetera.
Currently, all my tests use pytest.fail(..., pytrace=False), but this is far from ideal, as I have many assistants writing tests and not all of them are familiar with this necessity.

I made a fork that adds the --no-trace flag, opening the issue to submit a pull request.