Add a benchmark based on python -m pprint

[arhadthedev]

1. The moved code measures performance so it should belong here.
2. The relocation lifts off a question whether python -m pprint should have tests or not.

After merging, python/cpython#92546 needs to be closed.

[arhadthedev]

This creative PR is a receiving half of python/cpython#94613.

[ericsnowcurrently]

Thanks for working on this, Oleg! I've left some comments for you to consider.

[arhadthedev]

@ericsnowcurrently Thank you for your thorough feedback, I've addressed everything.

In addition, I've also moved a class docstring into a file comment, and reworded it together with authorship. I hope I've made it right.

Also I've added hasattr-based conditional running of _safe_repr to not fail CI for pre-3.10 CPython.

What did you have in mind relative to stdlib benchmarks? Note that we already have a bunch of micro-benchmarks for various parts of the stdlib, in distinct benchmarks.

Initially I've thought that a benchmark should be big, like Lib/test/*.py test files.

[ericsnowcurrently]

Thanks for the updates.