Add Primus benchmark collection to cybersecurity LLM resources

[Mic52M]

Pull Request: Add Primus benchmark collection to cybersecurity LLM resources

Summary

This PR contributes a new benchmark collection to the README.md in the Benchmark section:

• [Primus: A Benchmark for Evaluating LLMs in Cybersecurity Tasks](https://huggingface.co/collections/trendmicro-ailab/primus-67b1fd27052b802b4af9d243)

Description

The Primus collection, released by Trend Micro, includes multiple task-specific datasets focused on:

• Log parsing and classification
• Threat and anomaly detection
• Cybersecurity-focused reasoning and Q&A

The resource aligns well with the repository's goals and supports efforts to improve evaluation of LLMs on security-critical applications.

Let me know if you'd like adjustments in placement, formatting, or additional entries. Happy to iterate!