[SwastikTripathi]

extractUrls() and downloadListOfUrls() miss URLs on internationalized top-level domains, and cut punycode ones short:

extractUrls({ string: 'https://пример.рф/путь https://xn--e1afmkfd.xn--p1ai/x https://例子.中国/' });
// before: [ 'https://xn--e1afmkfd.xn' ]
// after:  [ 'https://пример.рф/путь', 'https://xn--e1afmkfd.xn--p1ai/x', 'https://例子.中国/' ]

URL_NO_COMMAS_REGEX and URL_WITH_COMMAS_REGEX already accept Unicode letters in the host labels (\p{L}), but the TLD was [a-z]{2,63}. So .рф, .中国 or .みんな never matched, and for a punycode TLD the regex stopped at xn and returned a URL with the wrong host.

The TLD is now xn--[a-z0-9-]{1,59} or [\p{L}\p{M}]{2,63}, with the punycode branch first so xn--p1ai is taken whole. I also added \p{M} next to \p{L} after the first character of the host and in the path, because scripts like Devanagari write vowel signs as combining marks: https://उदाहरण.भारत/page stopped at the first mark before. The first character of the host is unchanged.

The new test (5 URLs, both regexes) fails before the change. test/utils and the enqueue-links tests pass (220), and oxlint/oxfmt are clean.