This helper is used for extracting part of the issue / MR / whatever
description for use in the description meta tag:

1. To do that, we look at the source of the Markdown description.
2. We then strip out all HTML tags.
3. And then take the first 30 words.

Doing that can be really slow - especially as Markdown is supposed to be
treated as plain text. There are many better ways to do this, but the
immediate performance fix is to swap steps 2 and 3. This does mean that
the description may be less than 30 words (or even empty), but it is
much faster when the description is very long.