Fix SMTPUTF8 test to actually test non-ASCII email addresses

[matrixise]

Summary

Fixes the test_smtputf8_support test to actually test SMTPUTF8 (RFC 6531) functionality by using non-ASCII characters in email addresses, not just in message bodies.

Background

As pointed out in issue #5, the previous test claimed to test SMTPUTF8 support but only used ASCII email addresses (user@example.com, recipient@example.com) and UTF-8 characters in the message body. This doesn't test what SMTPUTF8 is designed for.

SMTPUTF8 (RFC 6531) allows non-ASCII characters in email addresses themselves, such as:

• prénom@société.fr (French accented characters)
• usuario@例え.jp (Japanese characters)

Changes

• Use actual non-ASCII email addresses: expéditeur@société.fr and destinataire@例え.jp
• Use EmailMessage: Switched from manual message construction to EmailMessage which uses modern email policy and properly handles SMTPUTF8
• Automatic SMTPUTF8: Use send_message() which automatically enables SMTPUTF8 when non-ASCII addresses are detected
• Verify envelope addresses: Check the X-MailFrom and X-RcptTo headers added by dsmtpd, which contain the actual SMTP envelope addresses
• Proper header decoding: Add MIME header decoding to correctly verify UTF-8 content

Test Results

All tests pass:

tests/test_server_integration.py::test_smtputf8_support PASSED

The test now properly verifies that:

1. SMTPUTF8 is announced in EHLO response
2. Non-ASCII email addresses are accepted
3. UTF-8 addresses are correctly preserved in envelope headers
4. UTF-8 content is preserved in message body

Related

Addresses feedback on issue #5