Drop deprecations

[bkamins]

Fixes #2344

I turn depwarns to errors where it is reasonably easy to do.

[bkamins]

One conclusion I came to is that:

• functions that we defined, but do not define any more should not have custom deprecations (like melt or aggregate) - as the point of removing deprecations is to free the name space of them in particular, so that the names can be used by users
• for getindex/setindex/setproperty I leave MethodError for unsupported functionality as writing proper error messages is very complex, the change that we remove deprecation is very old and MethodError actually prints good suggestions (i.e. to use two indices instead of only one)

[bkamins]

as a special case note that map for GroupedDataFrame CHANGES behavior (which is OK I think).

[bkamins]

OK - this is good to have a look at.

[quinnj]

LGTM; it's nice that a lot of these cases just turn into throw(ArgumentError(... with the same message, so the user is still getting helpful direction.

[bkamins]

a lot of these cases just turn into throw(ArgumentError(...

Yes, but I did not want to go too crazy with this, and left it where it was natural to leave and did not impact the code in a negative way (in some cases the rules of Julia dispatch made it difficult to preserve, and in 2 cases actually the depwarn was a dead code 😄, as later we made changes that made it impossible to hit that code path).

[nalimilan]

Maybe it would be worth leaving a comment to errors that are only there due to deprecations so that we can easily spot and remove them at some point in the future? Without comments, it could be hard to understand why we explicitly check for some unsupported patterns when you don't know the history of the package.

[nalimilan]

What do you think of the suggestion I made above?

Maybe it would be worth leaving a comment to errors that are only there due to deprecations so that we can easily spot and remove them at some point in the future? Without comments, it could be hard to understand why we explicitly check for some unsupported patterns when you don't know the history of the package.

[bkamins]

Ah - I do not know why I have missed this. I will add them.

[bkamins]

added

[bkamins]

Thank you!