DOC: Show constructor arguments for some classes in pd.series.offsets

[Dr-Irv]

This contributes to #52431

In that issue, the goal is to show the arguments for the constructors for the various offsets, which are all in pyx files.

The problem is that sphinx doesn't pick up the arguments for __init__() in those files. By adding the # cython: embedsignature=True at the top of the file, then it will pick it up for any class we document that has an explicit __init__() method. So, if you preview the docs from this MR, you will see it for classes like YearEnd, and BusinessHour because they have explicit __init__() methods. But you won't see it for YearBegin, for example, because it has no explicit __init__() method.

So to fix that, I illustrate how to fix it for Day, where I introduce an __init__() method that just calls the super().__init__(), and that makes sphinx then pick up the docs for pandas.tseries.offsets.Day

Also, I had to add the okexcept things to doc/source/user_guide/enhancingperf.rst because otherwise the docs would not build on Windows.

If this PR is accepted, then we can get the community to do the work of adding the various __init__() methods to the other documented offset classes and they can also add a Parameters section to those classes, which is what I did for Day as well.

[simonjayhawkins]

@Dr-Irv this and the following change have for some time caused pre-commit to fail for me locally.

great that this is being fixed here. Thanks.

Any idea why CI hasn't picked this up before now?

[Dr-Irv]

No, I was surprised as well.

[Dr-Irv]

/preview

[datapythonista]

No strong opinions about this, but if the examples need the okexcept on windows sounds like this change is breaking things on windows, no? I guess I'm missing something, if you can clarify. But it would be better to fix the underlying problem, I don't think we should add okexcept unless we document exceptions.

[Dr-Irv]

No strong opinions about this, but if the examples need the okexcept on windows sounds like this change is breaking things on windows, no? I guess I'm missing something, if you can clarify. But it would be better to fix the underlying problem, I don't think we should add okexcept unless we document exceptions.

No, the change is needed to just build the regular documentation (without this change) on Windows. It has something to do with how sphinx calls cython and that doesn't seem to work right on Windows. Not sure exactly what the problem is.