use rstar instead of bvh2d

[mockersf]

No description provided.

[codspeed-hq]

Merging this PR will degrade performance by 29.9%

⚡ 3 improved benchmarks
❌ 5 regressed benchmarks
✅ 33 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	baking	35.1 ms	18.9 ms	+85.74%
❌	is in mesh Vec2(135.0, 360.0)	2.8 µs	3.9 µs	-29.44%
❌	is not in mesh Vec2(726.0, 470.0)	4.5 µs	5.8 µs	-23%
❌	is not in mesh Vec2(297.0, 438.0)	4.4 µs	5.6 µs	-21.16%
❌	no path Vec2(297.0, 438.0)-Vec2(575.0, 410.0)	5.9 µs	7.1 µs	-17.18%
⚡	no path Vec2(458.0, 47.0)-Vec2(575.0, 410.0)	6 µs	5.4 µs	+11.17%
❌	is in mesh Vec2(728.0, 148.0)	3.1 µs	4.5 µs	-29.9%
⚡	no path Vec2(575.0, 410.0)-Vec2(458.0, 47.0)	6 µs	5.4 µs	+11.17%

---

_{Comparing drop-bvh2d-for-rstar (9baa48e) with main (ad45583)}