email: multiline encoded headers

Bug report

There seems to be a problem with multiline encoded headers, for example subject headers.

from email import message_from_string, policy
import base64

e = """From: <test@test.com>
To: <test@test.com>
Subject: =?UTF-8?B?0JfQsNGB0YLRgNCw0YXQvtCy0LrQsCAi0JDQu9C40LDQvdGGINCX0LTRgNCw0LLQ?=
 =?UTF-8?B?tSIg4oCTINCY0L3QtNC40LLQuNC00YPQsNC70LXQvSDQvdC+0LzQtdGAINC90LAg?=
 =?UTF-8?B?0YPQstC10LTQvtC80LvQtdC90LjQtQ==?=

test"""

e = message_from_string(e, policy=policy.default)

s1 = e["Subject"]

try:
    e.as_bytes()
except Exception:
    print("error")
else:
    print("ok")

s = (base64.b64decode("0JfQsNGB0YLRgNCw0YXQvtCy0LrQsCAi0JDQu9C40LDQvdGGINCX0LTRgNCw0LLQ") + base64.b64decode("tSIg4oCTINCY0L3QtNC40LLQuNC00YPQsNC70LXQvSDQvdC+0LzQtdGAINC90LAg")
    + base64.b64decode("0YPQstC10LTQvtC80LvQtdC90LjQtQ==")).decode("utf-8")

del e["Subject"]
e["Subject"] = s

s2 = e["Subject"]

try:
    e.as_bytes()
except Exception:
    print("error")
else:
    print("ok")

print(s1 == s2)

e = """From: <test@test.com>
To: <test@test.com>
Subject: =?UTF-8?B?0JfQsNGB0YLRgNCw0YXQvtCy0LrQsCAi0JDQu9C40LDQvdGGINCX0LTRgNCw0LLQtSIg4oCTINCY0L3QtNC40LLQuNC00YPQsNC70LXQvSDQvdC+0LzQtdGAINC90LAg0YPQstC10LTQvtC80LvQtdC90LjQtQ==?=

test"""

e = message_from_string(e, policy=policy.default)

s3 = e["Subject"]

try:
    e.as_bytes()
except Exception:
    print("error")
else:
    print("ok")

print(s2 == s3)

Your environment

Tested on Python 3.10.4 and 3.11.0a7 on Arch Linux

Run the provided reproduction with Python's email.message_from_string, policy.default, and as_bytes entry points. Compare s1, s2, and s3, including serialization failures, then trace the email package's handling of folded encoded Subject headers. Done means multiline and single-line encoded headers behave consistently without errors and the decoded values match.