Added an alternative way to initialize/load some models (for non-…

[shawl336]

Added an alternative way to initialize/load some models (for non-Android codes), tokens, hotwrods and keywords, that initializing/loading from memory buffers. The original usage of model initialization unchanged.
Basically, I just added additional "const char* $_buf_begin, const char* $_buf_end" vars ($ == encoder/decoder/joiner/model), if they are non-nullptr, sherpa will attempt to init the models from the memory starting from "$_buf_start" to "$_buf_end" in priority to the original filename strings.
Also, I added "tokens_buf_str, hotwords_buf_str and keywords_buf_str", if they are non-empty, sherpa will attempt to treat them as the content directly, and using istringstream rather than iftream.
the supported models are:
online models:
OnlineTransducerModel
OnlineParaformerModel
OnlineWenetCtcModel
OnlineZipformer2CtcModel
OnlineNeMoCtcModel

offline models:
  OfflineTransducerModel
  OfflineParaformerModel
  OfflineNemoEncDecCtcModel
  OfflineWhisperModel
  OfflineTdnnModel
  OfflineZipformerCtcModel
  OfflineWenetCtcModel
  OfflineSenseVoiceModel

keywordspotter models:
  KeywordSpotterTransducer

[csukuangfj]

Could you also add tests for your changes?

[csukuangfj]

Please add new fields to the end of existing fields.

Please don't change the order of exsiting fields.

[csukuangfj]

I suggest that you use

const char *encoder_buf;
int32_t encoder_buf_len;

[csukuangfj]

By the way, I suggest that you change one model config per pull-request and add test for the changes.

Otherwise, the pull-request is very large and you need to write lots of tests to cover your changes.

[csukuangfj]

Suggested change

	const char *encoder_buf_begin, *encoder_buf_end;
	const char *encoder_buf_begin = nullptr;
	const char *encoder_buf_end = nullptr;

1. Please follow our existing code style to define one variable per line.
2. Please initialize it to nullptr.

[csukuangfj]

Suggested change

	recognizer_config.model_config.transducer.encoder_buf_begin =
	SHERPA_ONNX_OR(config->model_config.transducer.encoder_buf_begin, nullptr);
	recognizer_config.model_config.transducer.encoder_buf_begin =
	config->model_config.transducer.encoder_buf_begin;

And please change other places.

[csukuangfj]

I suggest that you use

const char *encoder_buf;
int32_t encoder_buf_len;

[csukuangfj]

Please update the Validate() method to check that
when model_buf_begin is not nullptr, then model must be empty and vice versa.