This is a surprising effect of the utf8::upgrade/downgrade functions

What surprising effect? Their purpose is to convert a scalar's internal encoding, and I used them for that purpose.

If it helps clear up some confusion, change

   utf8::downgrade my $bin_dn = $bin;  # UTF8=0
   utf8::upgrade   my $bin_up = $bin;  # UTF8=1
[download]

to

   my $bin_dn = $bin;                    # UTF8=0
   chop my $bin_up = $bin . "\x{2660}";  # UTF8=1
[download]

Practical use for utf8::upgrade: Ensure "Unicode semantics" are used in regex matches. (But note that work is being done to remove such dependencies on this internal information.)

Practical use for utf8::downgrade: Ensure a string is a string of bytes (only contains chars 0-255), such as in Encode::decode and in Net::SFTP::Foreign::write.