[mkantautas]

Initial checklist

• I read the support docs
• I read the contributing guide
• I agree to follow the code of conduct
• I searched issues and discussions and couldn’t find anything or linked relevant results below
• I made sure the docs are up to date
• I included tests (or that’s not needed)

Description of changes

Character references in attribute values are decoded when parsing, but directiveToMarkdown writes & as it is.
So a value that contains text such as & or   changes when the markdown is parsed again:
{title: 'Write   here'} is serialized as title="Write   here", which parses back with a no-break space.
With remark and remark-directive, every formatting pass decodes one more level:

::tip{title="Write   to keep two words together"}

becomes title="Write   …" after one pass and a literal no-break space after the next.

This encodes the & of each character reference that would be decoded, as &.
It uses parse-entities with the same settings as the parser, so other ampersands (a & b, ?a=1&b=2, &param) are serialized as before.
That applies to quoted and unquoted values.
The # and . shortcuts are not used for such values, as & cannot be in a shortcut.