Previously, when the "invisible" flag was activated, we wouldn't print
anything except the "p: " prefix on a password line. This made it
difficult to verify the password, and if you for some reason had
multiple passwords, you wouldn't know which one to pick unless you knew
the order.

This at least gives you a visual cue about your password.

Instead of the convoluted hard-to-read `hide_password &&` nonsense, give
ourselves a way to express the condition in terms of the argument _not_
being one of our accepted options (only `-i`). This ensures that we
either capture the filename in the last argument or use STDIN.

Connect the `-i` argument flag to the renderer. If the arg was passed,
hide passwords, otherwise show them.

Change `args.first()` to `args.get(1)` because the actual first element
is the command executable.

Not happy with the condition I'm using here to grab the password file or
the STDIN `-`. It doesn't read well at all. It all does work, but this
part I want to rewrite to make it easier to read.

Turn on a `hide_password` boolean variable if an `-i` argument is passed
in (haven't tested this yet). This tells us whether the user wants us to
hide the password in the selection interface.

Instead of rendering the password string, hide it while allowing it to
be copied. It is identified by the "p: " prefix, but nothing follows
that.

This allows you to run `passextract` without having to worry about
others peering over your shoulder or about screen capture or recording
hacks.

I'll be adding a command line flag to force you to opt in to this
feature, while keeping the old functionality the same.

The last thing copied from Passextract will stick around in your
clipboard, potentially opening up your password to accidental pasting or
a clipboard exploit.

Pass deals with this in a nice way, by restoring your clipboard back to
what was copied before after a set time delay.

Here that same functionality is more difficult to achieve because the
Clipboard crate can only deal with strings. So if for example you've
copied a file, an image, a program's proprietary type, or some other
binary data, it can't be restored with the Clipboard crate. Pass is able
to do this because it uses the OS X `pbcopy`/`pbpaste` commands under
the hood, which do support binary data. We could do that here I suppose,
but it's easier and cross-platform to leverage a library.

My heavy-handed solution to the problem of clipboard insecurity is to
just overwrite the clipboard with an empty string when Passextract
quits. The solution is not ideal because it doesn't preserve your past
clipboard entry, and it forces you to keep the Passextract menu open
until you paste the copied entry, but it's better than nothing when it
comes to keeping the clipboard secure.

If there's an error writing to the clipboard, Passextract will refuse to
quit.

`g`: moves the selection to the first option
`G`: moves the selection to the last option

This allows for faster selection if there are more than 2–3 options.

Using `g` still allows `gg` to work since doubling the command doesn't
change its meaning.

Values that contained ": " would not be parsed correctly for copying to
the clipboard. If, for example, a password contained ": ", those
characters and any following them wouldn't be copied. This change allows
such passwords to be copied correctly.

Also include brief notices in the source files. Not a fan of the huge
blocks of text that the FSF recommends.