- The polish lexer was wrongly adapted from the american one: the code to trim
the prefix `$` char was kept instead of trimming the postfix `PLN`
- Anyway the regex isn't enforced to match until the end of input so trimming
postfix units is not needed
- And the trimming for `€` was wrong anyway, it assumed `€` is only 1 char long
in utf8
We'll need to factorise that better in the not-too-far future: there was already
a discrepancy between en and fr; and this won't scale if we keep adding
languages.