Matrix4: Rewrite makeOrthographic() for clarity

[WestLangley]

I thought it would be nice if the method could be deemed correct simply by inspection. :-)

The implementation is now similar to that of makePerspective().

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

	Before	After	Diff
WebGL	338.04 78.86	338.04 78.86	+0 B +0 B
WebGPU	559.1 154.77	559.1 154.77	+0 B +0 B
WebGPU Nodes	558.02 154.56	558.02 154.56	+0 B +0 B

🌳 Bundle size after tree-shaking

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

	Before	After	Diff
WebGL	469.29 113.56	469.31 113.52	+16 B -42 B
WebGPU	634.39 171.81	634.4 171.77	+16 B -44 B
WebGPU Nodes	589.52 161.12	589.54 161.08	+16 B -43 B

[Mugen87]

It seems the following unit test fails now:

not ok 1854 Maths > Frustum > setFromProjectionMatrix/makeOrthographic/containsPoint

[Mugen87]

@OrangeDoro Please stop writing these comments. They are considered as SPAM.

[Mugen87]

So I guess after #31456 a rebase should fix the CI?

[bhouston]

You've increased divides, which are the slowest operation, by 2x. From 3 to 6 no? FP divides are 2x to 5x slower than multiplications, but this is a micro-optimization.

The previous pattern of precalculating the inverses was the most efficient and pretty standard across math libraries.

[WestLangley]

This method is not called in a tight loop. Nor is it computationally expensive.

Compare makeOrthographic()in this PR with makePerspective(), and see how similar they are now.

Our policy has been to write for clarity first, and modify for speed if there is a demonstrated performance issue.

If this PR is merged, it would be fine with me to add the suggested micro-optimizations to both makeOrthographic()and makePerspective() in a subsequent PR, but I don't think doing so is necessary in this case.