Added support for Viking pre-tokenizer

[kustaaya]

Implemented suggested changes from discussion in pull request #7328 by @akx . Added pre-tokenizer support for Viking-models.

• Added tokenizer type for Viking-7B in convert-hf-to-gguf-update.py
• Added the chkhsh for Viking-7B in convert-hf-to-gguf.py
• Added LLAMA_VOCAB_PRE_TYPE_VIKING enum to llama.h
• Added pre-tokenizer regex for LLAMA_VOCAB_PRE_TYPE_VIKING to llama.cpp
• Ran ./tests/test-tokenizer-0 ./models/ggml-vocab-viking.gguf. Tests passed.

Tokenizer is the same for Viking-7B and its 13B and 33B siblings. Fixes issue #7309

• I have read the contributing guidelines
• Self-reported review complexity:
  • Low
  • Medium
  • High

[slaren]

The changes look correct me, but there are conflicts that need to be resolved before this can be merged. I think it is just because llama.h and llama.cpp were moved to a different directory.

[kustaaya]

The changes look correct me, but there are conflicts that need to be resolved before this can be merged. I think it is just because llama.h and llama.cpp were moved to a different directory.

Thanks! I updated the pull request to fix the conflicts.