Expand drop() overloads to encode more argument restriction and add type parameters to df.drop() signature.

[JanEricNitschke]

• Closes #xxxx (Replace xxxx with the Github issue number)
• Tests added: Please use assert_type() to assert the type of any return value

This is again to increase compatibility with pyright strict mode. I verified that the df.drop tests do not pass pyright strict without this change but do with it.

Also (maybe) closes #950.

This is done by adding a bunch of overloads that specify the allowed combinations in more detail.

For that i added a test that fails without this change and passes with it,

However i was not able to specify the full functionality using the overloads. Specifically, explicit "None" arguments are allowed by the signature, but cause errors at runtime. This is mostly because None is Hasheable.
This could be improved substantially in correctness and conciseness of the hints if it were possible to type hint the general signature using Non-None-Hasheable | None and then split these up in the overloads.

[JanEricNitschke]

One more thing that could be done would be to add overloads for the cases where the Nones are explicitely specified and mark that as deprecated and Never return. But i dont know how clear that would be. See: microsoft/pyright#8265

[Dr-Irv]

Thanks @JanEricNitschke