Cleanup/jsontable

[matthewpeterkort]

Summary

This PR updates Benchtop to align with Grip’s new tableId-first storage model by removing assumptions about label-based table resolution.

It ensures that benchmarks reflect the simplified storage architecture and remain accurate after recent Grip changes.

---

Key Changes

• Updated table resolution to use direct table IDs
• Removed dependencies on label indirection
• Simplified integration logic with Grip storage APIs
• Cleaned up outdated abstractions tied to schema-based lookups

---

Motivation

Grip recently simplified its storage model by eliminating label indirection. Benchtop needed to be updated to:

• Avoid stale assumptions about label resolution
• Ensure benchmark correctness
• Reflect real-world execution paths

This keeps benchmarking aligned with the current architecture.

---

Impact

Correctness

• Prevents mismatches between benchmark logic and runtime behavior
• Ensures consistent table resolution semantics

Maintainability

• Less coupling to evolving schema abstractions
• Cleaner integration with Grip internals

---

Notes

This PR does not introduce new benchmark features.
It is primarily a compatibility and alignment update following storage refactors in Grip.