SphereGeometry: Make pole vertices more robust.

[Mugen87]

Related issue: #11449

Description

While revisiting #11449, I have noticed that even alternative ray casting approaches struggle with SphereGeometry because of how the pole vertices are computed. Math.sin( Math.PI ) produces no exact zero but something like 1.2246467991473532e-16 which means the vertices never lie exactly on the local y-axis. By snapping the south-pole vertices exactly to the y-axis, the broken ray/triangle intersection test from #11449 (see https://jsfiddle.net/zraqkg2f/3/) starts to work with our present algorithm.

Even outside of ray casting, this simple fix makes the vertex data a bit more robust and exact.

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

	Before	After	Diff
WebGL	365.43 86.79	365.43 86.79	+0 B +0 B
WebGPU	661.52 182.6	661.52 182.6	+0 B +0 B
WebGPU Nodes	659.56 182.3	659.56 182.3	+0 B +0 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

	Before	After	Diff
WebGL	498.49 121.58	498.49 121.58	+0 B +0 B
WebGPU	732.76 197.41	732.76 197.43	-3 B +19 B
WebGPU Nodes	681.98 184.81	681.98 184.83	-3 B +20 B