datasets: refresh GameTora event dataset and scraper pipeline

[ETO111139]

This PR refreshes the Uma Musume dataset from GameTora and includes the scraping pipeline needed to keep it updated.

What is included

• integrate scrape_skills.py, scrape_events.py, and run_characters_supports_cpu.py into the repo
• add required Python dependencies for the scraper flow
• refresh datasets/in_game/skills.json
• refresh datasets/in_game/events.json
• rebuild datasets/in_game/event_catalog.json
• update web/public/events/*, web/public/icons/skills/*, and rebuilt web/dist/*
• fix a Windows cp1252 console print issue in build_catalog.py flow so catalog rebuild works on Windows too

Validation performed on March 20, 2026

• scraped skills successfully from the current GameTora skills JSON
• scraped supports and characters from the current GameTora sitemap
• rebuilt catalog successfully
• ran npm ci
• ran npm run build

Dataset status after refresh

• supports in events.json: 492
• trainees in events.json: 245
• scenarios in events.json: 2
• total root entries in events.json: 739
• entries in event_catalog.json: 9313

Coverage note

• current GameTora sitemap resolved to 526 support URLs and 245 character URLs
• final missing count after scrape: 0 supports, 0 trainees
• some support URLs in the sitemap are duplicates pointing to identical card/event data, so the final merged support count is 492 unique support entries rather than 526 raw URLs

This PR intentionally contains only the dataset refresh work and the scraper/build support needed for it.