Bug 24365: (bug 24217 follow-up) Fix encoding issue for translator script

I am not sure what happened here, but adding strict for TmplTokenizer.pm
breaks a test for translation.

It has been caught by a failing test:
 % prove xt/author/translatable-templates.t
will display lines "invalid multibyte sequence"

I did a diff between the generated files and found:
< msgid "Ã rbok"

Signed-off-by: Martin Renvoize <martin.renvoize@ptfs-europe.com>

misc/translator/xgettext.pl

@@ -366,7 +366,7 @@ usage_error('You cannot specify both --convert-from and --files-from')
 
 if (defined $output && $output ne '-') {
     print STDERR "$0: Opening output file \"$output\"\n" if $verbose_p;
-        open($OUTPUT, '>', $output) || die "$output: $!\n";
+        open($OUTPUT, '>:encoding(utf-8)', $output) || die "$output: $!\n";
 } else {
     print STDERR "$0: Outputting to STDOUT...\n" if $verbose_p;
     open($OUTPUT, ">&STDOUT");