Update POD and comments to be clearer about characters vs bytes

lib/RT/Action/SendEmail.pm

@@ -1129,7 +1129,8 @@ sub PseudoReference {
 
 =head2 SetHeaderAsEncoding($field_name, $charset_encoding)
 
-This routine converts the field into specified charset encoding.
+This routine converts the field into specified charset encoding, then
+applies the MIME-Header transfer encoding.
 
 =cut
 

lib/RT/I18N.pm

@@ -330,14 +330,8 @@ sub SetMIMEEntityToEncoding {
 =head2 DecodeMIMEWordsToUTF8 $raw
 
 An utility method which mimics MIME::Words::decode_mimewords, but only
-limited functionality.  This function returns an utf-8 string.
-
-It returns the decoded string, or the original string if it's not
-encoded.  Since the subroutine converts specified string into utf-8
-charset, it should not alter a subject written in English.
-
-Why not use MIME::Words directly?  Because it fails in RT when I
-tried.  Maybe it's ok now.
+limited functionality.  Despite its name, this function returns the
+bytes of the string, in UTF-8.
 
 =cut
 

lib/RT/Template.pm

@@ -375,10 +375,9 @@ sub IsOverride {
 Returns L<MIME::Entity> object parsed using L</Parse> method. Returns
 undef if last call to L</Parse> failed or never be called.
 
-Note that content of the template is UTF-8, but L<MIME::Parser> is not
-good at handling it and all data of the entity should be treated as
-octets and converted to perl strings using Encode::decode_utf8 or
-something else.
+Note that content of the template is characters, but the contents of all
+L<MIME::Entity> objects (including the one returned by this function,
+are bytes in UTF-8.
 
 =cut
 

t/web/rest-non-ascii-subject.t

@@ -4,7 +4,6 @@ use warnings;
 use RT::Test tests => 9;
 
 use Encode;
-# \x{XX} where XX is less than 255 is not treated as unicode code point
 my $subject = Encode::decode('latin1', "Sujet accentu\x{e9}");
 my $text = Encode::decode('latin1', "Contenu accentu\x{e9}");