Automated scoring of the Ambiguous Intentions Hostility Questionnaire using fine-tuned large language models