Please describe the problem:
In camel/camel-mime-utils.c:rfc2047_encode_word(), if the last unicode character
in the input string is non-ASCII, it is always encoded as a separated chunk.

Separating is valid in RFC2047, but separating in this case is not an intended
result.


Steps to reproduce:
A simple test program:
===============================================
#include <camel/camel-mime-utils.h>
#include <stdio.h>

int
main (int argc, char *argv[])
{
	char *encoded;
	char *header;

	header = "\xed\x95\x98\xed\x95\x98\xed\x95\x98";
	encoded = camel_header_encode_phrase (header);
	printf ("%s\n", encoded);
	g_free (encoded);
	header = "\xed\x95\x98\xed\x95\x98";
	encoded = camel_header_encode_phrase (header);
	printf ("%s\n", encoded);
	g_free (encoded);
}
===============================================
1. 
2. 
3. 


Actual results:
$ ./test-rfc2047
=?UTF-8?Q?=ED=95=98=ED=95=98?= =?UTF-8?Q?=ED=95=98?=
=?UTF-8?Q?=ED=95=98?= =?UTF-8?Q?=ED=95=98?=
$

Expected results:
=?UTF-8?Q?=ED=95=98=ED=95=98=ED=95=98?=
=?UTF-8?Q?=ED=95=98=ED=95=98?=


Does this happen every time?
Always

Other information: