Fix or rewrite scraper engine and scrapers

Many years have passed since its first release.

Known demozoo bug:
- it always select the first secondary mirror assuming that is from scene.org. This is wrong and should be carefully checked
- demozoo is missing the screenshots, it always downloads only the first one

Known scrape engine bugs:
- sometimes it sets "screenshots" to None: it should set that field to empty list if no screenshot detected
- it doesn't handle very well other extensions than gb, therefore, it may be a good idea to fix this thing, sometimes in the manifest there is "gbc" but the engine has already renamed the file in gb

Other improvements:
- change the general logic to become more flexible (e.g. select best source, include other extensions)
- write a basic test suite
- test from scratch other scrapers, since they may have become buggy due to change in the master scraper

Commencez par localiser le moteur de scraping, le scraper maître, les scrapers individuels et la gestion du manifeste mentionnés dans l’issue, puis examinez comment sont sélectionnées les sources, les captures d’écran et les extensions de fichier. Le travail est terminé lorsque les bugs Demozoo et du moteur listés sont corrigés, que les autres scrapers ont été vérifiés et qu’une suite de tests de base couvre le comportement attendu.

Rédigé par le modèle d'indexation à partir du texte de l'issue.