Resolve Inconsistency in Matrix3 and Matrix4 rotateY Implementations

[tlserver]

This PR addresses the issues highlighted in #261 and is related to the discussions in #262 and #300, as well as the historical context provided by #69.

In reference to the conversation in #262, two primary concerns were identified:

1. The change introduced test failures.
2. There was an inconsistency between the rotateY methods in the Matrix3 and Matrix4 classes.

Here's a summary of the remedial actions taken in this PR:

This PR is similar to #262. The only difference is that this PR also includes a negated rotation angle in the obb3 test.

For issue (1), the test expectations have been updated to reflect the correct signs, ensuring that all tests now pass.

Regarding issue (2), upon further investigation, it was found that the Matrix4 class already had the correct implementation since August 23, 2014. However, the Matrix3 class has not been updated to match this, leading to a longstanding inconsistency between the two. This PR rectifies the Matrix3 rotateY method to align with the Matrix4 implementation, bringing them into consistency.

  // Matrix3
  /// Turns the matrix into a rotation of [radians] around Y
  void setRotationY(double radians) {
    final c = math.cos(radians);
    final s = math.sin(radians);
    _m3storage[0] = c;
    _m3storage[1] = 0.0;
    _m3storage[2] = -s;
    _m3storage[3] = 0.0;
    _m3storage[4] = 1.0;
    _m3storage[5] = 0.0;
    _m3storage[6] = s;
    _m3storage[7] = 0.0;
    _m3storage[8] = c;
  }
  
  // Matrix4
  /// Sets the upper 3x3 to a rotation of [radians] around Y
  void setRotationY(double radians) {
    final c = math.cos(radians);
    final s = math.sin(radians);
    _m4storage[0] = c;
    _m4storage[1] = 0.0;
    _m4storage[2] = -s;
    _m4storage[4] = 0.0;
    _m4storage[5] = 1.0;
    _m4storage[6] = 0.0;
    _m4storage[8] = s;
    _m4storage[9] = 0.0;
    _m4storage[10] = c;
    _m4storage[3] = 0.0;
    _m4storage[7] = 0.0;
    _m4storage[11] = 0.0;
  }

This PR is similar to #262. The only difference is that this PR also includes a negated rotation angle in the obb3 test.

By merging this PR, we will resolve the inconsistencies and adhere to the mathematical definitions of rotation matrices, as expected by the users of the vector_math.dart library.

I look forward to your feedback and to further improving the library.

[spydon]

Lgtm, I think this is quite important to land now when 3D is more and more used in Flutter

[johnmccutchan]

Do we have the same tests in Matrix4?

[johnmccutchan]

Please update this PR to include a change log entry and then I'll merge it.

Thanks for the fixes!

[coveralls]

coverage: 26.33%. remained the same
when pulling 5a190ff on tlserver:issue261
into 3706feb on google:master.

[tlserver]

@johnmccutchan

Thank you for merging the PR! Could you please provide an estimated timeline for when version 2.1.5 will be published to include this fix?
Also, #261 can be closed.

Thanks!