MCM: fix auto-detection of vLLM binary cache

[fulvius31]

has_artifact_compile_range_with_triton() only checked for a triton/
subdirectory, which cannot exist when the artifact is a packed binary
file. Recognize binary artifact_compile_range* files as valid vLLM
cache indicators so detect_cache_mode() returns 'vllm' instead of
falling through to 'triton'.

Also:

• sync requirements.txt with pyproject.toml (typer[all], structlog)
• silence pylint R0903 on Pydantic data models
• disable pylint import-error for declared but not-installed deps

[Copilot]

Pull request overview

This PR fixes auto-detection of vLLM binary cache by updating the cache detection logic to recognize binary artifact files in addition to directory-based artifacts. The detection function previously only checked for triton/ subdirectories, causing binary artifacts to be misidentified as triton caches.

Changes:

• Updated _has_artifact_compile_range_with_triton() to detect both binary artifact files and unpacked artifact directories
• Synchronized requirements.txt with pyproject.toml for typer and structlog dependencies
• Added pylint configuration to suppress warnings for Pydantic models and import errors

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
mcm/requirements.txt	Synced dependency versions with pyproject.toml, pinned vllm version
mcm/pyproject.toml	Removed triton dependency, pinned vllm version, added pylint configuration
mcm/model_cache_manager/utils/utils.py	Fixed cache detection to recognize binary artifact files
mcm/model_cache_manager/tests/utils/test_cache_mode_detection.py	Added test case for binary artifact cache detection
mcm/model_cache_manager/data/kernel_validator.py	Added pylint disable comments for Pydantic model classes

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[fulvius31]

cc @maryamtahhan