i didn't double check if Runemax has made it into
libc.h.  i think it makes sense, even for 16-bit runes since
Bgetc returns an int.  and cases like this where a rune
value is calculated as an int to protect against overflow.
this patch depends on Runemax.

1.  fix output bug.  for character 0, the output was truncated
because %C formats a 0 character, terminating the string.
also defend against illegal utf values.

2.  add -b baseoff to correct the character values for subfonts
that don't start at character 0.  while this is a hack (we should
parse font(6) files instead so that each subfont could be
discontinuous with this previous), it does make viewing a single
subfont possible with much less work.