Ah, I see why you were thinking of uniqueness... Maybe something like this, combining the idea of max() and uniqueness via hash:

...
# group arrays by person+id
my %persons;
for my $aref (@$array_ref_x) {
    my $uniq_key = "$aref->[0] $aref->[1]";
    push @{$persons{$uniq_key}}, $aref;
}
    
my $all_val = [];
for my $aref_person (values %persons) {
    push @$all_val, my $tmp = [];
    for my $aref (@$aref_person) {
        @$tmp[0..1] = @$aref[0..1];
        for my $i (2..$#$aref) {
            $tmp->[$i] = 0 unless defined $tmp->[$i];
            $tmp->[$i] = $aref->[$i] if $aref->[$i] > $tmp->[$i];
        }
    }
}

print Dumper $all_val;

__END__
$VAR1 = [
          [
            'John Day',
            '456789',
            2011,
            '50.00',
            817,
            4,
            22,
            90
          ],
          [
            'ALICE WONDER',
            '9876543',
            2009,
            '15.00',
            717,
            20,
            5,
            11
          ],
          [
            'Mary Ann',
            '123988',
            2002,
            '10.00',
            119,
            5,
            3,
            0
          ]
        ];
[download]