chore: load py_library from rules_python

[rickeylev]

The native.py_* symbols are deprecated, newer versions of Bazel don't have them,
and newer versions of rules_python have mixed support for them.

Switch to loading them from rules_python instead, which are supported.

[hawkinsp]

This looks ok to me but let's see what the CI says.

(We're currently on a slightly older rules_python, but an update should land soon.)

[rickeylev]

I think rules_python 0.39 is currently being used, which should be new enough.

If there's any issues, it's likely to be in the copybara transforms. native.py_* was eliminated within Google, so if native.py is showing up on github, then it would have to be a c copybara transform (or oss-specific file) that was doing it.

[hawkinsp]

I think rules_python 0.39 is currently being used, which should be new enough.

If there's any issues, it's likely to be in the copybara transforms. native.py_* was eliminated within Google, so if native.py is showing up on github, then it would have to be a c copybara transform (or oss-specific file) that was doing it.

Nope, we should be good. We don't transform this file's contents (and instead just point to a different file with the same API).

[hawkinsp]

You want me to merge this? Mark it as non-draft?

[rickeylev]

I've marked as ready for review. I had just left it in draft to lookup how to copybara transform it manually, but if that's a minor concern, then ready to merge