Feat/final release

[MahrukhBaig]

What does this PR do?

[One sentence: what does this PR add, fix, or change?]

Which module(s) does it touch?

[List the files changed and why each was changed]

How did you test it?

[What tests cover this? Run: pytest <test_file> and paste the result summary]

Edge cases handled:

[What inputs could break this? How did you handle them?]

Checklist:

• Tests added or updated
• Docstrings on all new public functions
• No unused imports
• README updated if public API changed
• Lead Engineer has reviewed and approved before submitting to director

[Khubaib8281]

tests/test_performance.py

• Could we mark this as a separate performance/integration test? A hard 2s threshold may be flaky across different CI runners.

tests/test_factual_grounding.py

• Could we avoid downloading the model during the regular test suite? Perhaps mark these as integration tests and run them in an environment where the model is pre-cached.

tests/test_cli.py

• The test name says this verifies repeated --criteria, but only one criterion is passed here. Could we pass it twice and assert both are included in the output?

tests/test_evaluator.py

• Minor: since CRITERIA_REGISTRY is global, could we use a fixture to restore its state after these tests? This would make the tests less sensitive to execution order.

README.md

• Since omitting criteria runs all registered criteria, could we mention that adding future criteria may also increase the evaluation cost/latency?