libtecla should not use legacy ctype.h macros
The older macro versions of ctype.h tests are poor, as they lack good support for certain characters, especially in foreign character sets. See bug #330 for an example. (The legacy version only supports 8 bits of type data.)
The better solution would be to either use functional versions, or -D_XPG4 to use slightly better macro versions. Of course long term, the code should support multi-byte character sets, perhaps using wide character versions of the macros (iswtype() or iswprint()) to gain maximum support for multibyte locales.
This should probably be coordinated with the upstream source for libtecla, as well.