RNNLM model support lm_num_thread and lm_provider setting (#173)

* rnnlm model inference supports num_threads setting

* rnnlm params decouple num_thread and provider with Transducer.

* fix python csrc bug which offline-lm-config.cc and online-lm-config.cc arguments problem

* lm_num_threads and lm_provider set default values

---------

Co-authored-by: cuidongcai1035 <cuidongcai1035@wezhuiyi.com>

sherpa-onnx/csrc/offline-lm-config.h

@@ -16,11 +16,17 @@ struct OfflineLMConfig {
 
   // LM scale
   float scale = 0.5;
+  int32_t lm_num_threads = 1;
+  std::string lm_provider = "cpu";
 
   OfflineLMConfig() = default;
 
-  OfflineLMConfig(const std::string &model, float scale)
-      : model(model), scale(scale) {}
+  OfflineLMConfig(const std::string &model, float scale, int32_t lm_num_threads,
+                  const std::string &lm_provider)
+      : model(model),
+        scale(scale),
+        lm_num_threads(lm_num_threads),
+        lm_provider(lm_provider) {}
 
   void Register(ParseOptions *po);
   bool Validate() const;