MNT: deprecate epoch2num/num2epoch

[jklymak]

PR Summary

epoch2num and num2epoch convert to and from matplotlib times from unix times.

We do not use these functions anywhere internally, and their functionality is provided by datetime or datetime64 modules. There is really no reason for users to be converting directly from unix times to Matplotlib times.

This was deprecated before when we made the epoch change, but that broke pandas who were using this. They no longer are using this, so I think it is safe to deprecate for 3.5.

PR Checklist

• Has pytest style unit tests (and pytest passes).
• Is Flake 8 compliant (run flake8 on changed files to check).
• New features are documented, with examples if plot related.
• Documentation is sphinx and numpydoc compliant (the docs should build without error).
• Conforms to Matplotlib style conventions (install flake8-docstrings and run flake8 --docstring-convention=all).
• New features have an entry in doc/users/next_whats_new/ (follow instructions in README.rst there).
• API changes documented in doc/api/next_api_changes/ (follow instructions in README.rst there).

[jklymak]

I'm conflicted about julian2num. It is obscure enough that many folks might be counting on it.

[jklymak]

I'll ping for this simple one...

[timhoffm]

What's the first relase of pandas that does not depend on this? I think we need to give it "some time" to settle.

I'm fine with deprecation the julian functions. I consider them a is too specifc and not part of the core functionality that we want to provide. The few potential users of this should be able to copy the code or get the conversion from somewhere else.

[timhoffm]

I think we leave tests for deprecated features in place and run them with _api.suppress_matplotlib_deprecation_warning():.

[jklymak]

It was merged almost a year ago: pandas-dev/pandas#35393 milestoned 1.1, released Jul 28, 2020. They are currently on 1.2.4. I would guess that if someone updates Matplotlib they would also update pandas? Worse that happens is someone gets a deprecation warning for a while if they use a new Matplotlib and an old pandas.

[timhoffm]

That's ok then.

[timhoffm]

Modulo one comment