How can the proto token match the B sym rather than the A sym?

proto-tokens.raku

use v6;

grammar Test {
    token TOP { <x>+ % \h+ $<rest>=(.*) }
    proto token x {*}
    token x:sym<A> { \S+ [\h \S+]* }
    token x:sym<B> {
        "[" <special-token> "]" <.ws> <important-message>
    }
    token special-token { "SPECIAL TOKEN" }
    token important-message { \S+ }
}

my $dat = q (59.5   Some Term   [SPECIAL TOKEN] Important!!  Yet another term);
say Test.parse($dat);

# ｢59.5   Some Term   [SPECIAL TOKEN] Important!!  Yet another term｣
#  x => ｢59.5｣
#  x => ｢Some Term｣
#  x => ｢[SPECIAL TOKEN] Important!!｣
#  x => ｢Yet another term｣
#  rest => ｢｣