A (but copied many) typo of char-mapping tables

While I looked into a .map file in src/backend/utils/mb/Unicode, I
notice of a typo in it.

> static const pg_mb_radix_tree euc_jp_from_unicode_tree =
> {
> ..
> 0x0000, /* offset of table for 1-byte inputs */
> ...
> 0x0040, /* offset of table for 2-byte inputs */
> ...
> 0x02c3, /* offset of table for 3-byte inputs */
> ...
!> 0x0000, /* offset of table for 3-byte inputs */
> 0x00, /* b4_1_lower */
> 0x00, /* b4_1_upper */
> ...
> };

Yeah, the line above prefixed by '!' is apparently a typo of "4-byte
inputs", which comes from a typo in convutils.pm.

Fortunately make maintainer-clean; make all in the directory results
in no other differences so we can apply the attached patch to fix the
all propagated typos.

I don't think no backpatch is needed.

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center