tests: policy operator edge cases in test_policy_evaluator

[CodewithJha]

👋 New to open source?
Thanks for considering contributing to Mutiny.
This issue is intentionally beginner-friendly.
If you get stuck, ask questions in the comments.
We're happy to help.

Context

Policy evaluation is deterministic in Core. tests/unit/test_policy_evaluator.py already covers deny / require / forbid / numeric / context paths. Empty args and nested-path edges still have room without changing semantics.

Goal

Add focused unit cases for empty-args and/or nested path edge behavior, mirroring existing test class style.

Why it matters

Edge coverage protects contributors who extend operators later and documents expected behavior for the cheatsheet / examples.

Acceptance criteria

• Add focused unit cases (empty tool args and/or nested context / arg paths)
• uv run pytest tests/unit/test_policy_evaluator.py -q passes
• No production behavior change unless a clear bug is found (call that out in the PR)
• PR description links this issue

Suggested files

• tests/unit/test_policy_evaluator.py
• Read-only: operator / evaluator code under packages/mutiny_core/

Out of scope

• New operators, changing constraint meaning, Hosted API tests

Hints

• Mirror patterns in TestRequireArgsRefundLimit, TestContextConstraints, TestInvalidConstraints.
• Prefer one small class or a few methods over a large rewrite.
• If behavior is already correct, the PR is tests only — that’s success.

Difficulty

S · tests only

Docs

• Catalog: GOOD_FIRST_ISSUES.md #15