support bpe tokenizer in convert

[ftgreat]

Our released Aquila models used bpe tokenizer, so in convert.py we just add one branch for preprocessing bpe tokenizer vocab into sentencepiece in order to use following modules like inference or int4. we have make sure all encoding ids are all the same and have no impact other modules.

Could you please review this pr, thanks.
Related issue: #2093

[howard0su]

Can you provide test instruction so that I can verify the change?

[ftgreat]

Can you provide test instruction so that I can verify the change?

instruction:
python convert.py models/7B --vocab-only --outfile models/aquila-vocab.bin --vocabtype bpe

requirements:
put vocab.json in models dir, vocab.json from Aquila-tokenizer https://github.com/FlagAI-Open/FlagAI/blob/master/examples/Aquila/Aquila-tokenizer-hf/vocab.json

[klosax]

Note: Using an llama model with gpt2 tokenizer will be fully supported in the new ggml file format. ggml-org/ggml#302

[ftgreat]

Note: Using an llama model with gpt2 tokenizer will be fully supported in the new ggml file format. ggerganov/ggml#302

Could you please give me the support schedule?
And how to add our released models, thanks.