Expose JPH::PlaneShape via JPC_PlaneShapeSettings

[Solidor777]

Problem

JoltC does not expose JPH::PlaneShape (the infinite-half-space collider; "negative half space is solid"). Downstream consumers that want a real halfspace primitive — for example mapping a Halfspace { normal } shape from a higher-level physics abstraction — have no choice but to approximate with a thin static box, which loses the authored normal direction and forces an arbitrary thickness offset.

Fix

Add JPC_PlaneShapeSettings + matching _default / _Create exports following the same pattern as JPC_BoxShapeSettings, JPC_SphereShapeSettings, etc.

typedef struct JPC_PlaneShapeSettings {
    // ShapeSettings
    uint64_t UserData;

    // PlaneShapeSettings
    // TODO: Material
    JPC_Vec3 Normal;
    float Constant;
    float HalfExtent;
} JPC_PlaneShapeSettings;

JPC_API void JPC_PlaneShapeSettings_default(JPC_PlaneShapeSettings* object);
JPC_API bool JPC_PlaneShapeSettings_Create(const JPC_PlaneShapeSettings* self, JPC_Shape** outShape, JPC_String** outError);

The C-side struct carries Normal + Constant rather than a JPC_Plane wrapper to avoid introducing a new binding type for a single use site; the C++ side constructs the JPH::Plane via its Plane(Vec3 normal, float constant) ctor. Material is stubbed with a TODO comment, consistent with every other *ShapeSettings in the codebase.

HalfExtent defaults to JPH::PlaneShapeSettings::cDefaultHalfExtent (1000.0f), matching JPH's own default.

Constraint

PlaneShape MUST be static or kinematic per JPH (PlaneShape::MustBeStatic() == true). Caller's responsibility, same contract as JoltPhysics/Samples/Tests/Shapes/PlaneShapeTest.cpp upstream.

Validation

Downstream verification: Titan engine's titan-jolt-3d plugin uses this binding to back its BodyShape::Halfspace { normal }. New integration test (halfspace_authors_non_y_normal) raycasts a non-+Y halfspace and asserts:

• ray distance lands on the authored plane (5 m for a 5-m offset along the plane normal)
• hit normal returned by Jolt's raycast matches the authored normal (dot product > 0.95 with the expected direction)

The same test against the prior thin-box approximation returned axis-aligned normals on the box faces, never the authored plane normal — concrete evidence that the new path actually exercises PlaneShape::GetSurfaceNormal.

Test plan

• Builds against current main on Windows host (msvc).
• JPC_PlaneShapeSettings_Create returns a valid JPC_Shape* for a +Y normal at the origin.
• Downstream raycast against the resulting body returns the authored normal.
• Material parameter — out of scope for this PR (left as TODO like every other shape).

[LPGhatguy]

Hello and thank you!

Can you remove the titan comments from the code? Then it's good to merge!

[Solidor777]

Done — removed both TITAN PATCH comment blocks; section headers now match the rest of the file.

[LPGhatguy]

Merci beaucoup!