Support building GPU-capable sherpa-onnx on Linux aarch64. (#1500)

Thanks to @Peakyxh for providing pre-built onnxruntime libraries 
with CUDA support for Linux aarch64.

Tested on Jetson nano b01

sherpa-onnx/csrc/offline-paraformer-model.cc

@@ -56,7 +56,7 @@ class OfflineParaformerModel::Impl {
 
   const std::vector<float> &InverseStdDev() const { return inv_stddev_; }
 
-  OrtAllocator *Allocator() const { return allocator_; }
+  OrtAllocator *Allocator() { return allocator_; }
 
  private:
   void Init(void *model_data, size_t model_data_length) {