Avoid reading fields for empty projections

[joseph-isaacs]

Superseded

This PR is intentionally empty and closed. The COUNT(*) fix was folded into the PRs that own each part of the behavior:

• Add plan optimizer rules and push expressions #9196: zero-reference expression reduction in ExpressionRowIdxRule
• Add plan-native scan execution #9240: childless-Pack row-count endpoint during split discovery
• Prune plan scans with zoned statistics #9244: end-to-end zero-segment-read regression

The all-in-one benchmark PR #9342 contains the same distributed history.

[codspeed-hq]

Merging this PR will improve performance by 19.44%

⚡ 1 improved benchmark
✅ 1961 untouched benchmarks
⏩ 89 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	cold_misaligned[(64, 256)]	5.2 ms	4.4 ms	+19.44%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ji/plan-count-star (a836fa0) with ji/vx-plan-ci (cecbc8f)}

Footnotes

1. 89 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[joseph-isaacs]

Folded this fix into the stack at the owning boundaries: #9166 handles the zero-partition expression pushdown, #9240 handles the childless Pack split endpoint, and #9244 contains the end-to-end zero-read regression. This branch now matches #9341 exactly, so this PR has no remaining diff.