getting kind of crazy common now and being employed even when totally
unnecessary (e.g. encoding strings that contain just plain ASCII):
- normalize message header fields, when possible
- normalize UTF-8 encoded characters, when possible (e.g. special punctuation
  chars)

This allows text filters (e.g. subject.can, name.can) to work on MIME-encoded
header fields and notifications about received e-mails are legible to humans.

Encoded-words that contain actual non-ASCII/CP437 chars (e.g. foreign symbols,
emojis) are left as encoded-words to be dealt with by whatever displays the
message header.

Special handling of folded normalized field values was necessary because
"White space between adjacent 'encoded-word's is not displayed." (per RFC 2047)