Respect tokenizer.ggml.add_bos_token value when tokenizing (#4040)

* gguf-py: gguf-dump: Respect --no-tensor flag in JSON mode.

* Respect add_bos_token GGUF metadata value

* gguf-py: Try to fix SpecialVocab giving up too easily for the Nth time

common/common.h

@@ -200,6 +200,10 @@ std::string llama_detokenize_bpe(
                          llama_context * ctx,
         const std::vector<llama_token> & tokens);
 
+// Uses the value from the model metadata if possible, otherwise
+// defaults to true when model type is SPM, otherwise false.
+bool llama_should_add_bos_token(const llama_model * model);
+
 //
 // YAML utils
 //