Python: Add initial TableScan implementation

[Fokko]

This adds an implementation of TableScan that is an alternative to the one in #6131. This doesn't implement plan_files, it is just to demonstrate a possible scan API:

scan = table.scan(
    row_filter=In("id", [5, 6, 7]),
    selected_fields=("id", "data"),
    snapshot_id=1234567890
)

[rdblue]

Looks good.

[rdblue]

This looks good to me, except that it left out the scan refinement methods. I think those should be included so you can use the same refinement strategy that the JVM API supports.

[Fokko]

Also threw in some tests 👍🏻

[rdblue]

Thanks, @Fokko!