perf: optimize V3 search and staging

[skyoo2003]

Description

Optimize V3 search hot paths and reduce Redis staging round trips. MemoryEfficient ASCII searches use a direct root index, FindBatch calls the serving engine directly, and V3 chunk/manifest staging is pipelined.

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactoring (no functional changes)
• Test update

Checklist

• Tests pass (make all)
• Vet/make vet
• Linting passes (make lint)
• Build succeeds (make build)
• Documentation updated if needed
• Changelog fragment added (internal performance optimization and tests only)
• Commit messages follow guidelines

Additional Notes

Valkey 8.1.10 Docker benchmark on 10k keywords showed MemoryEfficient ASCII engine median improvements of 21.9% for Find and 27.1% for FindIndex; V3 shared-text p95 improved 17.9% for Find and 22.7% for FindBatch. Multibyte paths were unchanged within benchmark variance.

Benchmark setup: valkey/valkey:8-alpine, localhost port mapping, Go 1.27.1, 3 repeated runs. Full repository verification passed: make all.