Cleanup: dict word printing

[erkyrath]

There's some messy logic which could be improved.

print_dict_word() decodes Z-characters and tries to show them in escaped form (@{3c1} or @oe). We could reasonably soften this to allow native printing of characters in the -Cn set, or UTF8 printing in -Cu mode. This is already partially done, in fact; but we're going through a function called word_to_ascii() which makes life confusing.

(word_to_ascii() is misnamed: it can output Latin-1 characters.) (Also, it's Z-code only and the name should reflect that!)

While we're at it, why is dict word printing handled in two places? Why do the printing functions have the annoying dual d_show_buf logic? Rethink this entire area.