[VL] Pushdown scan and filter to remote Storage

### Description

One common issue on cloud is the remote storage throughput. Currently we use rowgroup skipping to reduce the data size.
The plan is to pushdown the scan and filter as substrait plan to remote storage, get the data in arrow format and input into pipeline.

No files, tests, or entry points are identified. Start by locating the existing rowgroup-skipping path, then determine how scan and filter operations could be represented as a Substrait plan for remote storage and how Arrow-format data would re-enter the pipeline; done requires an agreed design and working end-to-end pushdown.

Written by the indexing model from the issue text.