fix: route clip_manager logging through module logger

[shezmic]

What changed

Two logging consistency fixes in clip_manager.py, both surgical — no logic changes.

1. ClipEntry.find_assets() — logging.warning() → logger.warning()

The module defines logger = logging.getLogger(__name__) at line 26, but two call sites in find_assets() bypass it, writing directly to the root logger via logging.warning():

• Empty AlphaHint directory warning (was line 153)
• AlphaHint directory with no valid media files warning (was line 164)

These now use the module logger, so they respect the caller's log configuration and show the correct module name (clip_manager) in log output.

2. scan_clips() — print() → logger

The INVALID OR SKIPPED CLIPS summary and the all-valid confirmation used raw print(), which:

• Bypasses log level filtering entirely
• Is invisible to any log handler when clip_manager is used as a library
• Cannot be suppressed without redirecting stdout

Replaced with logger.warning() per invalid entry and logger.info() for the all-valid case.

Why it was needed

When clip_manager is imported and called programmatically (e.g. from corridorkey_cli.py), print() output goes directly to stdout and cannot be filtered or redirected by the caller's logging setup. The logging.warning() calls had the same issue — they wrote to the root logger, not the module logger, making them harder to trace and impossible to suppress per-module.

How to verify

uv run pytest          # 221 passed, 2 skipped
uv run ruff check clip_manager.py   # All checks passed

No behaviour change in interactive use — log output appears identically at WARNING/INFO level when the root handler is configured (as corridorkey_cli.py does via _configure_environment()).