Merging new audio functionality into Octave

[Mike Miller]

I'm interested to hear how the maintainers feel about merging the new
audio functionality that was developed for GSoC into Octave. The work
that was done includes audio file I/O and sound card I/O.

I think that the new modules are reasonably functional, but could
definitely benefit from more testers, especially on Mac and Windows.
In particular, I would be excited to see it merged in before the next
major release, but I understand if the consensus is that it's not
ready or that there's not enough time to make it ready before the
release.

There is already a public clone of Octave available with the new audio
classes and functions integrated into the build:

  https://bitbucket.org/bucket_brigade/octave-with-sound

Here's what I know of that definitely needs to be fixed in that
development branch before I'd consider it ready to merge into the
official repo:

  * Commit message cleanup (wrap long lines, set hg.user to real name and email)
  * Add --without-portaudio and --without-sndfile options to configure
  * Remove %!test blocks that use sound hardware, no good for make check
  * Mention in NEWS (does this deserve a paragraph dedicated to audio?)

I'd also recommend

  * Rewrite legacy play/record functions to use new class-based functions
  * Rewrite legacy wavread/wavwrite functions to use audioread/audiowrite

A --enable-audio option could be added if we want to treat this as an
experimental disabled-by-default feature for one release cycle.

The new functions depend on new dependency libs, portaudio and
sndfile. Both are licensed compatibly, cross-platform, and already
part of MXE.

All comments and testers welcome, thanks,

-- 
mike