speakup: convert screen reading to 16bit characters

This adds 16bit character support to most of the screen reading by
extending characters to u16 throughout the code.

Non-latin1 characters are assumed to be alphabetic type for now.

non-latin1 vt_notifier_call-provided characters are not ignored any
more, and the 16bit character returned by get_char is not truncated any
more. For simplicity, speak_char still only supports latin1 characters.
Its direct mode however does support 16bit characters, so in practice
this will not be a limitation, non-latin1 languages will be handled by
the synthesizer. spelling words does not support direct mode yet, for
simplicity for now it will ignore 16bit characters.

For simplicity again, speakup messages are left in latin1 for now.

Some coding style is fixed along the way.

Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Reviewed-by: Okash Khawaja <okash.khawaja@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>