LINEAR24 userland format in audio(4) - do we really want it?

[nia <nia%NetBSD.org@localhost>]

I believe this should not be enabled, and that applications
should be trained to write 32-bit linear samples instead.

The reason being that it's very confusing how exactly
24-bit samples should be encoded, and different applications
and implementations have different ideas.

In OSSv4, S24 means "24-bit samples packed into 32-bit
words". It has a separate AFMT_S24_PACKED for 24-bit
samples packed into 24-bit words.  FreeBSD does not appear to
support the PACKED format, only using 4 bytes per sample.

Solaris does support the PACKED format when using OSSv4,
and appears to use it internally.  For the Sun API, the
intention is to use 3 bytes per sample.

For Sun .au files (which our audio interface is closely
mapped to), in Solaris 24-bit explicitly means 3 bytes
per sample.

I think the confusing situation means that a lot of
applications will be broken.