fix: treat non-breaking space as a separator in links (#66)

[patchwright]

Problem

Fixes #66. A non-breaking space (U+00A0) next to an e-mail address or URL is swallowed into the link instead of separating it. For example test@example.com\u{a0} (NBSP after) extends the e-mail link into the following text, and https://example.com\u{a0}now pulls the NBSP and now into the URL. Confirmed on main HEAD b663d4e.

Root cause

All three scanners admit U+00A0 as a valid character because it sits in the non-ASCII range they accept for internationalized text: local_atom_allowed (_ => c >= '\u{80}') in email.rs, the ALPHA arm '\u{80}'..=char::MAX in domains.rs, and the URL delimiter set in url.rs which stops at \u{9F}, one code point below NBSP. So NBSP never reaches a separator/break path even though it is whitespace per Unicode.

Fix

Treat U+00A0 as a separator in all three scanners: exclude it from local_atom_allowed, break the authority scan on it, and add it to the URL "never part of a URL" set. 3 source lines changed, no API change; other non-ASCII characters (e.g. U+03F8) are unaffected.

How to test

cargo test --test email non_breaking_space_does_not_join_email
cargo test --test url non_breaking_space_does_not_join_url

Both new tests fail on main and pass with this change; the full suite stays green (91 passed).

Backward compatibility

No breaking changes. NBSP is whitespace and is not a valid character in e-mail local-parts, domain labels, or URLs (RFC 3986/3987, RFC 5321/6531), so no previously valid link changes; links that previously ran through an NBSP now terminate at it.

---

Assisted-by: Claude (code generation, reviewed and tested locally)