[Type] Quat : Add more unit tests and two bugfixes

[fredroy]

While adding more extensive unit tests, 2 bug were detected by Claude:

1. createFromRotationVector(Real, Real, Real) at line 540: Changed phi >= 1.0e-5 to phi <= 1.0e-5. The condition was inverted compared to the Vec3 overload, causing it to return identity for any real rotation and
  divide-by-zero/NaN for zero input.
2. slerp(const Quat&, Real) at line 390-394: Changed sin(t * angle) / cos(t * angle) to sin(t * angle / 2) / cos(t * angle / 2). Quaternions encode rotations using half-angles, but quatToAxis returns the full rotation angle. The
  reconstruction was missing the half-angle division, producing incorrect interpolation for all t != 0

The one on createFromRotationVector looks good to me but as for slerp, Quat maths are a bit obscure, so I asked an other one (🤖)

Why the other AI is right

In your quatToAxis function, you explicitly calculate the full rotation:
angle = 2 * half_theta;

When you go to rebuild the interpolated quaternion in slerp, you are trying to create a rotation of magnitude t×angle. According to the formula above, the w component (the scalar) must be the cosine of half that magnitude.

Without that division by 2, your interpolation would rotate twice as fast as it should. If you were trying to rotate 90° over 1 second, your code would actually complete the 90° turn in 0.5 seconds and keep going to 180°.

[ci-build][with-all-tests]

---

By submitting this pull request, I acknowledge that
I have read, understand, and agree SOFA Developer Certificate of Origin (DCO).

---

Reviewers will merge this pull-request only if

• it builds with SUCCESS for all platforms on the CI.
• it does not generate new warnings.
• it does not generate new unit test failures.
• it does not generate new scene test failures.
• it does not break API compatibility.
• it is more than 1 week old (or has fast-merge label).