fix(editor): unload ghost preview model when path changes

setGhostPreview reused modelId 59999 for every preview, but loadModel
returns true without doing anything when the ID is already cached. So
selecting a new NPC kept the old ghost model in GPU memory and createInstance
used the stale model. Added M2Renderer::unloadModel public API and call it
from clearGhostPreview.

include/rendering/m2_renderer.hpp

@@ -287,6 +287,8 @@ public:
 
     bool hasModel(uint32_t modelId) const;
     bool loadModel(const pipeline::M2Model& model, uint32_t modelId);
+    /** Force-remove a model and all its GPU resources. Caller must ensure no instances reference it. */
+    void unloadModel(uint32_t modelId);
     /** Mark a loaded model as a spell effect (full-brightness particles, no collision). */
     void markModelAsSpellEffect(uint32_t modelId);