Add initialization phase to v2 file reader (and re-enable pushdown filtering)

[westonpace]

Pushdown filtering was prototyped but never fully implemented. One of the challenges is that pushdown filtering requires an "initialization" phase which could be combined as part of opening the file or separate. The reasoning for making it separate is that we might not know, at file open, which fields we need to access and use in filters.

We also need to introduce some kind of metadata cache that schedulers / decoders have access to. This would be populated on initialization (or first read) and could be reused for future reads.