Fix: Ignore query strings and fragments in path depth scoring - #2444

Summary

Fixes #2443.

PathDepthScorer counts slashes in query strings and fragments as path levels. With optimal depth 2, /docs/guide?next=/a/b/c scores 0.25 instead of 1.0, which can change deep-crawl priority. A URL without a path can also pick up a slash inside a query parameter.

Remove literal query/fragment suffixes before locating and counting the path. Preserve weighting, caching, repeated/trailing slash handling and encoded path characters.

List of files changed and why

• crawl4ai/deep_crawling/scorers.py: one behavioral line strips URL suffixes; scoped Black formatting also removes whitespace-only lines in the affected class and adds the missing final newline.
• tests/unit/test_path_depth_scorer.py: 12 assertion cases cover suffixes, pathless URLs, encoded separators, weights and repeated calls.

How Has This Been Tested?

• On unchanged develop: 7 cases fail and 5 controls pass.
• After the fix: all 12 new assertion cases pass. The existing general scorer diagnostic also reports all scorer/composite checks passed (13 pytest cases total).
• Black on the new test and the changed source line, Ruff on the new test, and git diff --check pass.
• Real editable checkout, Python 3.12.14, pytest 9.1.1. No browser/network mocks. Full suite, browser crawling and model tests were not run.

Checklist

• Added regression tests and ran the focused checks listed above.
• Human self-review: this patch and its tests were prepared and reviewed with Codex; no human review is claimed.
• Documentation: no API or configuration change.