Re: try-completion is inconsistent when completion-ignore-case=t

[Jonathan Yavner]

RMS> When the string needs to be changed for other reasons, Emacs adjusts
RMS> the case.  When the string as given is a match, Emacs returns it
RMS> unchanged.

But "test" is not actually a match.  It's "Test" that's the match.  We're
fuzzing the difference via completion-ignore-case.  Suppose you're typing
in incremental text and the program is displaying the best match so far:

(1) t         => Test
(2) te        => Test
(3) tes       => Test
(4) test      => test
(5) testo     => TestOne
(6) teston    => TestOne
(7) testone   => TestOne

It looks weird to have the initial lowercase `t' only on line (4).