Fix NPE in Geometry's morph state (#2919)

[DieterDePaepe]

Geometry allocates its morphState array lazily. setMorphState(float[]) and getMorphState() both create it on demand, but the name-based overloads setMorphState(String, float) and getMorphState(String) indexed the field directly, so calling either one first threw NullPointerException:

java.lang.NullPointerException: Cannot store to float array because
"this.morphState" is null

Nothing else populates the array. MorphControl only reads it, and the control itself is attached by GltfLoader only when the asset carries an animation channel targeting "weights". An application driving morph targets from its own code -- expression blending, a slider, anything procedural -- therefore has an unallocated array precisely when it reaches for the name-based setter, which is the natural one to use since it is the only API that resolves a morph by name.

Route both overloads through getMorphState(), which already allocates on demand, so the by-name and by-array paths behave alike. The lookup guard is unchanged: an unknown name still writes nothing and reads -1.

[Copilot]

Pull request overview

Fixes lazy morph-state initialization for name-based accessors.

Changes:

• Routes name-based reads and writes through the allocating getter.
• Adds regression tests for allocation and unknown names.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
Geometry.java	Prevents NPEs in name-based morph access.
GeometryMorphStateTest.java	Tests lazy allocation and lookup behavior.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.