Follow-up: address Copilot code quality findings from PR #356

[jordanpadams]

Summary

Two code quality issues were identified by Copilot during review of PR #356 (Skip crawling of excluded directories during path resolution) that should be addressed as follow-up work.

Issues to Address

1. resolve_ingress_paths() silently discards an explicitly passed empty list

File: src/pds/ingress/util/path_util.py

Problem: The initialization guard resolved_paths = resolved_paths or list() treats an explicitly passed empty list [] as falsy and allocates a new list, silently discarding the caller's reference.

Fix:

# Before
resolved_paths = resolved_paths or list()

# After
if resolved_paths is None:
    resolved_paths = list()

2. get_path_progress_bar() caches PATH_BAR globally regardless of call arguments

File: src/pds/ingress/util/progress_util.py

Problem: After PR #356 is merged, get_path_progress_bar() accepts includes, excludes, and follow_symlinks parameters to compute the correct total. However, subsequent calls with different parameters still return the cached PATH_BAR initialized from the first call, potentially producing incorrect totals (progress bar overrunning 100%) and causing test bleed between test cases that call the function with different filter settings.

Fix: Guard the cached value so it is only reused when called with the same parameters, or enforce and document single-call semantics with a clear reset mechanism.

Acceptance Criteria

• resolve_ingress_paths() initializes resolved_paths only when None is passed, not when [] is passed
• get_path_progress_bar() computes the correct total when called with different filter parameters, or raises a clear error if called more than once with conflicting settings
• Existing and new unit tests pass (pytest tests/)
• No regressions in path resolution behavior

Context

Identified by GitHub Copilot during review of PR #356.