Add BWIKI lineup archives and avatar scraping

[bupt-lxc]

Summary

• Add lineup_scraper.py to archive BWIKI player lineups into data/lineups.json and data/lineups.csv, with incremental refresh based on list-page update dates.
• Extend rocom_scraper.py to parse upload-lineup avatar images from the live BWIKI page, download them through the existing URL pipeline, and support --avatars-only / --skip-avatars modes.
• Refresh generated data indexes, including data/urls.csv avatar entries and updated sprite/skill datasets.

Fixes / Safety

• Preserve existing lineup output if all detail fetches fail, and record failed lineup/image URLs for retry.
• Skip already-downloaded avatar images via the existing _add_url cache/path behavior.
• Keep avatar assets separate from large sprite art under data/images/avatars, avoiding data/images/sprites mixing.
• Add tracked avatar parser fixture so tests do not depend on ignored data/*.html samples.

Test Plan

• python -m pytest tests/test_rocom_scraper.py tests/test_lineup_scraper.py -q -> 29 passed
• python -m py_compile rocom_scraper.py lineup_scraper.py tests/test_rocom_scraper.py tests/test_lineup_scraper.py
• python rocom_scraper.py --avatars-only -> parsed and saved 200 avatar URL rows, 0 failures

[AofeiLi-code]

Thank you for your contribution!