[Python] Add read_parquet, to_parquet and to_csv

[Tishj]

This PR adds some methods ported from pandas to our python API.

It also deprecates the old write_csv method, making it instead an alias to to_csv.
Also write_parquet is added as an alias to to_parquet

[Mytherin]

Looks like there are some non-std::moves in this PR - could you patch them out?

[carlopi]

Somehow related, I see this pattern where there are a bunch of read_csv, read_parquet, read_???, would potentially not make sense to have at some point a generic file ingesting functionality that takes as parameter the kind of file?

Something like read(fileLocation, 'csv') or read(fileLocation, 'json', {schema: "..."}) that also takes an optional third parameter with eventual options. And equivalent for output.

And then on the other side say you implement read_yourformat, it would be nice to be able to register that (either compile time or INSTALL time) into the generic read() handler.

I was discussing something similar regarding [de]compress with @samansmink, unsure whether this has value enough that (after some more thinking) this can be considered (probably at the library level so that can be then exposed also to SQL / Python / other bindings.

[Tishj]

@carlopi I think there is value in that, but definitely in addition to methods like these.
Because we already have table_function relation, which can also be used to run the read_csv_auto table function, these methods added in this PR are just syntactic sugar to make it easier to use these functions.

[Mytherin]

Thanks!