Client: Increase the maximum GPU buffer size to 256 MB

This is the default WebGPU limit, so it should be safe to use. The existing one is too small to render large maps with normal visualization enabled (such as schg_dun01 or nif_fild01) - there's just too many vertices since each vertex normal necessitates three more for the visualization itself.

Core/NativeClient/WebGPU/GPU.lua

@@ -10,6 +10,7 @@ local ffi_string = ffi.string
 local GPU = {
 	MAX_VERTEX_COUNT = 200000, -- Should be configurable (later)
 	MAX_TEXTURE_ARRAY_SIZE = 32,
+	MAX_BUFFER_SIZE = 256 * 1024 * 1024,
 }
 
 -- The FFI bindings don't provide enums for native extensions yet (requires a fix in the runtime)
@@ -116,7 +117,7 @@ function GPU:RequestLogicalDevice(adapter, options)
 				maxVertexAttributes = 4, -- Vertex positions, vertex colors, diffuse texture UVs, normals
 				maxVertexBuffers = 4, -- Vertex positions, vertex colors, diffuse texture UVs, normals
 				maxInterStageShaderComponents = 8, -- #(vec3f color, vec2f diffuseTextureCoords, float alpha), normal(vec3f)
-				maxBufferSize = self.MAX_VERTEX_COUNT * 5 * ffi.sizeof("float"), -- #(vec3f position, vec2f transform)
+				maxBufferSize = GPU.MAX_BUFFER_SIZE, -- DEFAULT
 				maxVertexBufferArrayStride = 20, -- #(Rml::Vertex)
 				maxBindGroups = 3, -- Camera, material, transforms
 				maxUniformBuffersPerShaderStage = 1, -- Camera properties (increase for material, soon?)