Bug 215341

This was actually fixed by the link detection rewrite merged into master months ago, but I forgot to notice and close this until Nicolás pointed it out.

Changelog:

* Detection of URLs and email addresses in chat text views to turn them into
  clickable links has been much improved. An incomplete overview of notable
  cases:
  - Unicode characters in URLs are now handled properly, enabling support
    for example for international domain names.
  - Protocol-less links not starting in "www.", such as the short URLs pop-
    ular these days ("bit.ly/foo" and similar), are now recognized.
  - Arbitrary protocols (e.g. "http://" or "message://") are now recognized;
    previously only a small hand-picked and insufficient set was.
  - The check that avoids balanced a pair of parentheses around an URL be-
    coming part of the link now works for more than one level of balanced
    parentheses.
  - Aside from balanced pairs of parentheses, also square and other forms
    of brackets are now handled properly.
  - Trailing question marks no longer become part of the link.
  - URLs using uncommon schemas, e.g. Apple message:// URLs or Wolfram
    Alpha URLs, are now handled properly.
  - Numerous improvements to email link handling: User names containing the
    plus sign are now handled correctly, opening email links from the URL
    Catcher works now and others.
  - The URL detection for the "Open URL" sub-menu in the list entry context
    menu in Channel List tabs now uses the same infrastructure as link detec-
    tion elsewhere rather than separate code, making it massively better com-
    pared to previous versions.
  - Average speed of link detection has improved slightly.