associated UTF-8 handling: have unget_input() filter out invalid wide
characters, put back the result of a word sequence instead of returning
it directly, don't erroneously filter out keystrokes if they're neither
extended keypad values nor ASCII characters, and add a few comment
tweaks


git-svn-id: svn://svn.savannah.gnu.org/nano/trunk/nano@2166 35c25a1d-7b9e-4130-9fde-d3aeb78583b8