DEPR: keep_date_col, nested parse_dates in read_csv

[jbrockmendel]

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

[mroeschke]

Thanks @jbrockmendel

[sm-Fifteen]

What is the rationale for deprecating datetime parsing on combined columns? That's something I personally use a lot (CSV being what it is, date, time of day and timezone offset are often stored in separate columns), and I've not seen equivalent code examples that would achieve the same thing at parsing time.

[mroeschke]

I've not seen equivalent code examples that would achieve the same thing at parsing time.

Internally pandas didn't do this at parsing time. read_csv would combine parsed columns and use to_datetime essentially with these options. You can achieve the equivalent by using .str.cat with to_datetime

[sm-Fifteen]

@mroeschke: Using to_datetime is fine for a flat dataframe, but it gets increasingly complicated as indexes and multi-indexes get tossed into the mix. No longer being able to specify those things in the read_csv config so I don't need to think about these afterwards wouldn't be the end of the world, but it's an extra bit of boilerplate index manipulation for some very common situations that I don't currently need to think about, and may pose problems for inexperienced users.