bug#7014: 23.2; `C-h c' command doesn't show the info when invoked from the minibuffer

[Drew Adams]

> > Try the following:
> > 1.- Start emacs (anyhow).
> > 2.- Type `C-x C-f'.
> > 3.- Now, from the minibuffer, type `C-h c' --> Here, Emacs 
> prompts me
> > for a key to describe. OK.
> > 4.- Type `C-n' (for example). --> Now Emacs should show me the
> > description of the `C-n' key (*), but that doesn't happens. Instead,
> > after a couple of seconds, the minibuffer comes back with the prompt
> > of the `find-file' command.
> 
> Fixed in trunk.

I have not seen the fix yet, but I certainly hope that the currently active
keymaps are used for this.

In particular, with the example given, the doc for `C-n' should list the command
it is bound to as being `exit-minibuffer', since the question about its binding
is being asked with a minibuffer keymap current, and that is the binding of
`C-n' in the minibuffer.

This is important.  Emacs should always faithfully answer questions about itself
and its current state.  When in the minibuffer, users should be able to ask and
get answers to questions about the current, minibuffer, state.

If for some reason some particular context/code needs to instead interrogate the
keymaps that were current before the initial command (`C-x C-f in this case), it
can do that (explicitly).

But a simple `C-h c' when the minibuffer is active should (must) DTRT and give
the currently active binding.  Anything else would be counter the Emacs
philosophy of dynamic introspection.  Emacs is not only self-documenting editor,
it documents itself in real-time.  And nothing about Emacs is more important
than that.