fix: support system columns in dataset.take* operations

[hamersaw]

Previously, "take*" operations did not support _rowid, _rowoffset, _row_created_at_version, and _row_last_updated_at_version. In this PR we add support for all of these columns.

We preserve these system columns through the initial schema projection so that they can be used to populate the correct flags when building the ProjectionPlan and PhysicalProjection structs.

• _rowid / _rowaddr: persisting these through to ProjectionPlan fields was enough to make them work
• _rowoffset: required additionally (1) stripping ROW_OFFSET field from ProjectionPlan requested_output_expr and (2) manually injecting column using AddRowOffsetExec (after exposing some methods publicly)
• _row_created_at_version / _row_last_updated_at_version: required piping through flags to Fragment readers.

Addresses #5615.

[github-actions]

ACTION NEEDED
Lance follows the Conventional Commits specification for release automation.

The PR title and description are used as the merge commit message. Please update your PR title and description to match the specification.

For details on the error please inspect the "PR Title Check" action.

[codecov]

Codecov Report

❌ Patch coverage is 85.10638% with 7 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
rust/lance-core/src/datatypes/schema.rs	84.09%	5 Missing and 2 partials ⚠️

📢 Thoughts on this report? Let us know!

[Xuanwo]

Thank you @hamersaw for working on this! Only have a question.

[hamersaw]

I'm pretty unsure about this (feels hacky). I first attempted to inject the ROW_OFFSET field into the schema within project_batch and inject the AddRowOffsetExec into the ExecutionPlan but was getting "index out of bounds" with len 2. I think this may be because we expect the ROW_ADDR column (required for ROW_OFFSET compute) but we manually inject it for take* operations.

My other idea is to strip this field from the requested_output_expr in the TakeBuilder when we convert into the ProjectionPlan. I'm unsure of the implications there.