fix: batch code-health fixes (#671, #790, #894, #909)

[microsasa]

Code-health batch

Five commits addressing open code-health issues and guideline cleanup.

Commits

1. Remove unreachable dead branch in _build_completed_summary — closes [aw][code health] Small cleanups: unreachable dead branch in _build_completed_summary + empty PRODUCT_VISION.md #671

   • Deleted if idx < len(events) else None guard that could never trigger
   • Removed test that manufactured the impossible scenario

2. Use default_factory=lambda for complex generic Pydantic fields — closes [aw][code health] Two Pydantic fields suppress reportUnknownVariableType instead of using default_factory=lambda: [] #790

   • toolRequests and shutdown_cycles fields changed from default_factory=list + # pyright: ignore → default_factory=lambda: []
   • Updated CODING_GUIDELINES with lambda guidance for complex generics

3. Remove arbitrary line-count threshold and dead-code guideline

   • Removed ~250-line module threshold (meaningless metric)
   • Removed Defensive Programming section that encouraged unreachable guard clauses

4. Fix stale cli.py docstring + allow getattr/hasattr in tests — closes [aw][code health] Minor guideline-compliance cleanup: stale docstring and banned hasattr/getattr in tests #894

   • Added missing vscode command to cli.py module docstring
   • Added test exception to getattr/hasattr guideline (introspection in tests is fine)

5. Detect ambiguous partial pricing matches and fall back — closes [aw][code health] _cached_lookup partial pricing match is nondeterministic when multiple keys tie on match_len, silently mis [Content truncated due to length] #909

   • When _cached_lookup finds multiple keys tying on overlap length, falls back to unknown-model pricing (1.0×) and logs a warning
   • Previously silently used whichever key was first in dict insertion order

Testing

All checks pass: ruff lint/format, pyright, bandit, unit tests (99% coverage), e2e tests (86 passed).

[Copilot]

Pull request overview

Code-health batch addressing several open issues across parsing, pricing, models, CLI docs, and contributor guidelines to reduce dead code, improve typing hygiene, and make pricing lookups deterministic/safer.

Changes:

• Simplifies session shutdown summary building by removing an unreachable guard and dropping the corresponding impossible test.
• Updates Pydantic Field defaults for complex generics to use default_factory=lambda: [] and documents the convention in CODING_GUIDELINES.
• Adjusts pricing partial-match logic to detect ambiguous ties, warn, and fall back to unknown-model pricing; updates tests accordingly.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/copilot_usage/pricing.py	Adds tie detection in partial-match pricing lookup and falls back to unknown-model pricing with a warning.
tests/copilot_usage/test_pricing.py	Updates/expands tests to assert tie fallback behavior and warning emission.
src/copilot_usage/parser.py	Removes unreachable shutdown index guard when building completed summaries.
tests/copilot_usage/test_parser.py	Removes the test that fabricated an impossible out-of-bounds shutdown index scenario.
src/copilot_usage/models.py	Replaces default_factory=list + pyright ignores with default_factory=lambda: [] for complex generic list fields.
src/copilot_usage/cli.py	Fixes module docstring to include the vscode command.
.github/CODING_GUIDELINES.md	Updates guidance on Pydantic defaults, removes dead code-health guidance, and permits getattr/hasattr in tests for introspection.