[A21.5] Switch away from ISO 2022 for the ISO-8859 character sets.

[Stephen J. Turnbull]

APPROVE 21.5

Aidan Kehoe writes:

 >  > This must be done *immediately*, as otherwise there is no way to read
 >  > files that have inadvertantly been saved with ISO-2022 extensions.
 > 
 > C-u C-x C-f /path/to/file RET iso-2022-8 RET .

I don't believe this will correctly read a file saved with the
iso-8859-2 coding system that also contains Latin-3 or Hebrew
characters, because G1 is initialized to latin-iso8859-1, not to
latin-iso8859-2.  Yes, that used to happen, and happen silently (it
was the original motivation for latin-unity).  I have one such file in
$HOME (ok, it was created for the purpose of verifying the bug) from
2001.  I'd be very surprised if nobody else has any.

 > GNU have it. It's not neurosurgery.

I know; I just want it done before you check in something that is not
backwards compatible.  Of course it's too late for that....