perf(zql): reduce allocations with frozen sentinel and object reuse

[Karavil]

Note: This PR is part of an upstream contribution effort from the Goblins team (@goblinshq). Co-authored with Claude by Anthropic.

Summary

Reduce GC pressure on the IVM push hot path with two allocation optimizations.

Motivation

During IVM push operations, every source change is propagated to all connected pipelines. For each connection, the current code creates fresh Node and Change objects with new relationships: {} literals. In a workload with 135 IVM pipelines processing ~200 row changes, this creates thousands of short-lived objects per push cycle, increasing GC pressure and pause time.

Similarly, generateWithOverlayInner creates {row, relationships: {}} node objects for every row during fetch, adding more allocation overhead on the fetch hot path.

Changes

• Add EMPTY_RELATIONSHIPS frozen sentinel at module scope, replacing per-node {} allocation in generateWithOverlayInner (3 sites) and genPush (via reuse objects)
• Pre-allocate reusable Change objects in genPush (reuseNode, reuseOldNode, reuseAddRemove, reuseEdit) and mutate row fields before yielding, instead of creating new objects per connection per push

Safety

Object reuse in genPush is safe because filterPush consumers are synchronous within the generator chain. Each yield* filterPush(...) completes fully before the next loop iteration mutates the reused objects. There is no concurrent access.

Expected Performance Impact

For 135 IVM pipelines each processing ~200 rows, this eliminates thousands of short-lived {} and Change object allocations per push cycle. The frozen sentinel also enables V8 to share the same hidden class across all nodes, improving inline cache hit rates for downstream consumers that read node.relationships.

Testing

• All 56 existing memory-source tests pass
• All 154 existing source tests pass

---

Stack Order

This PR is part of a stacked series of IVM performance optimizations. Merge in order:

1. perf(zql): reduce allocations with frozen sentinel and object reuse #5609 (this PR) - Allocation reduction
2. perf(zql): cache primary index key, pkConstraint, and index lookups #5610 - Index caching
3. perf(zql): comparator fast paths + compareBounds null fix #5611 - Comparator fast paths
4. perf(zql): fuse fetch pipeline, add PK fast path, reduce allocations #5612 - Fetch pipeline fusion

Independent PRs (no conflicts): #5607 (BTree iterators), #5608 (Join optimizations)

[vercel]

Someone is attempting to deploy a commit to the Rocicorp Team on Vercel.

A member of the Team first needs to authorize it.

[vercel]

Deployment failed with the following error:

The provided GitHub repository does not contain the requested branch or commit reference. Please ensure the repository is not empty.