DOC: fix EX02 docstring errors for three plotting docstrings

[phershbe]

Some work towards issue #51236
The issue is being worked on by several people and has more docstrings remaining to do.

Makes docstrings pass for:
pandas.plotting.andrews_curves
pandas.plotting.autocorrelation_plot
pandas.plotting.lag_plot

[phershbe]

@MarcoGorelli This is currently a draft because I'm unsure about something.

I got these docstrings to pass by changing instances of AxesSubplot in the Expected outputs to what the Got outputs said in a few places, I don't know if this is correct though. The other options would be that there are mistakes in the docstrings in the ways the methods are called or in the parameters passed in or something, or that there are mistakes in the methods themselves.

I don't know what I would be searching for in the alternative possibilities that I mentioned and maybe what I did was correct anyway and nothing else is necessary, so I thought I would make a draft and ask.

[MarcoGorelli]

Hey - looks like these are failing in CI when run as doctests

https://github.com/pandas-dev/pandas/actions/runs/4256504265/jobs/7405473685

But then they do pass when run as part of the docstring validation job...I don't know why that might be, I'll take a look

[phershbe]

@MarcoGorelli Oh, maybe it's because infer_dtype is not passing ./ci/code_checks.sh docstrings. I think it was accidentally broken in another pull request on this issue, or maybe broken for some other reason, or maybe it was broken by this pull request. I was pretty sure before that that mistake wasn't from this pull request, but I'll check soon here. I figured that we would clean it up later but didn't think about that it would be necessary for the validation job to clear.

Or maybe that's not it? In the failures, it's giving the same errors that were there before my changes, which were the reasons that these ended up on the --ignore_functions list to begin with. It's also listing pandas.plotting.parallel_coordinates and pandas.plotting.radviz as failures, even though those haven't been worked on yet and are still on the --ignore_functions list.

[MarcoGorelli]

I think we should just doctest+SKIP these plotting ones, see

#51614 (review)

[phershbe]

@MarcoGorelli uh oh, I'm not very good at Git yet and was having a lot of trouble fetching from the main upstream repository and then pushing my changes without conflicts, and I deleted the commits hoping to start fresh not realizing that I would close the issue by doing that. Obviously I would have preferred to have kept the pull request here, but it closed. I'm working on opening a new pull request now, sorry for making a mess of this.