Don't skip non-existent whitespace

The current code's skip pattern matches zero whitespace characters when
skipping whitespace.  The result in the lalrpop lexer is that a
zero-length match on no whitespace is found as a possible match.  In the
common case where a longer match (the next token) was found, lalrpop
discards the shorter match and returns the longer.  If no longer match
was found, lalrpop has special handling to note that if only a
zero-length match was found, that means no valid ones were.  So matching
zero-length whitespace doesn't actually cause a problem, but it's also
useless.

So write the expression we intend, which is to skip actual whitespace.

src/parser.lalrpop

@@ -330,7 +330,7 @@ IPv6: CascadeString = {
 
 // lexing precedence
 match {
-	r"[[:space:]]*" => { },
+	r"[[:space:]]+" => { },
 	r"//[^\n\r]*[\n\r]*" => { },
 	r"([[:digit:]]{1,3}\.){4}" => IPv4Regex,
 	"::1" => IPv6Regex, // TODO