llama : the WPM vocabs use the CLS token as BOS

[ggerganov]

llama.cpp/src/llama-vocab.cpp

Lines 1529 to 1535 in 9d5c711

	case LLAMA_VOCAB_TYPE_WPM:
	{
	if (add_special) {
	GGML_ASSERT(vocab.special_cls_id != -1);
	output.push_back(vocab.special_cls_id);
	}

Not sure if this separation of BOS and CLS tokens is really necessary. The alternative would be to remove the notion of CLS tokens from the codebase and work only with BOS tokens - seems much simpler. The problem would be if there are models that have both BOS and CLS tokens that are distinct?