fix: only use eos_token_id as pad_token_id if int

[dvrogozh]

LLama 3 has a list of values as eos_token_id:

  "['<|end_of_text|>', '<|eom_id|>', '<|eot_id|>']"

This breaks tokenizer since it expects single value. This commit uses tokenizer.eos_token_id instead in such a case.

Fixes: #2440

CC: @Narsil @zucchini-nlp @ladi-pomsar

[zucchini-nlp]

@dvrogozh Cool, looks good to me. We also need a review from TGI team