perf: Hash Grid Broadphase + Allocation Reduction

[eonarheim]

===:clipboard: PR Checklist :clipboard:===

• 📌 issue exists in github for these changes
• 🔬 existing tests still pass
• 🙈 code conforms to the style guide
• 📐 new tests written and passing / old tests updated with new scenario(s)
• 📄 changelog entry added (or not needed)

==================

This PR explores a different data structure for handling large numbers of colliders in the scene at a time efficiently.

• New features

  • Query colliders directly on the PhysicsWorld

    const scene = ...;
    const colliders = scene.physics.query(ex.BoundingBox.fromDimensions(...));

• New SparseHashGrid data structure

  • Used as the new default for the collision broadphase which is faster performing than the dynamic try
  • Used in the PointerSystem to improve pointer dispatch when there are a lot of entities

• More allocation reduction

  • New pool type RentalPool
  • Transform refactoring to remove
  • AffineMatrix
  • Graphics context state/Transform stack hot path allocations pooling

• Perf improvements to CircleCollider bounds calculations

• Switch from iterators to c-style loops which bring more speed

  • Entity component iteration
  • EntityManager iteration
  • EventEmitters
  • GraphicsSystem entity iteration
  • PointerSystem entity iteration

[cloudflare-workers-and-pages]

Deploying excaliburjs with    Cloudflare Pages

Latest commit:	8de93ce
Status:	✅  Deploy successful!
Preview URL:	https://316cee6d.excaliburjs.pages.dev
Branch Preview URL:	https://perf-collision-broadphase.excaliburjs.pages.dev

View logs