Attempts to fix failures of the Windows Audio Session API (WASAPI) integration which reportedly break the very capture and playback functionality.

This doesn't change anything semantically since the exceptions could
have already been thrown. They were just removed from the signatures in
an earlier commit and that might have been an omission during
refactoring.

Fixes a problem in the resolution of the devices by name (which could lead to the use of audio devices from an incorrect audio system).

Adds additional debug/error information output in the Windows Audio Session API (WASAPI) integration.

The new ZRTP cache handling implements two main topics.

The first modification reduces the number of write requests to the external
storage and simplifies overall handling of a ZRTP cache entry in the stack.

The second modification implements a new handling of the RS2 data. This will
be a topic in an upcoming revised ZRTP specification.

Caused by the isAccessible check in AudioSystem failing.
The fix is just to remove this check (as discussed with lubo over IM)

Fixes a problem with quickly displaying video preview can leave the camera open, adds listeners earlier to be sure we will get ui events to stop player.

Ignores the 'RTP pacing threshold' setting if it's set to over 1000 (as opposed to failing with an exception).

Attempts to improve the RTP packet queue/the audio jitter buffer with resepect to recovery when the queue/buffer is full. Works on bringing uniformity to the media processing threads created by FMJ and libjitsi as part of an effort to resolve quality degradation made very audible by jitsi-videobridge.

Improves handler for DTMF events: when sending several DTMF tones in a short frame, the notification sound is no more clipped but played entirely. The net.java.sip.communicator.service.protocol.maximalRtpDtmfToneDuration property is added to define a maximal duration for the RFC 2833/4733 DTMF tones.

Conflicts:
	trunk/lib/installer-exclude/libjitsi.jar

Fix for the issue when the encoder and the rtpConnectors are in separate VideoMediaDeviceSession instances and the encoder doesn't receive the key frame requests from other peers.

Improves setting of the stream direction when processing jingle content-accept messages. This fixes a problem with video conferences where video streams sometimes 'freeze' if a non-focus starts to send video (while the focus is not sending video)

Implements an alternative to PortAudio on Windows which uses Windows Audio Session API (WASAPI) and related Core Audio APIs such as Multimedia Device (MMDevice) API.

Changes the way FEC is handled in the Opus decoded. Makes it output a single Buffer containing both the data decoded from FEC and the packet's 'regular' data. Until now, it output two separate Buffers.

Adds an option to set the video bitrate. Patches provided by Benoît Laniel on the dev@ mailing list (under the subject of '[PATCH] Set default max video bandwidth to 128')

Corrects the RTP DTMF buffering management for the following case: when a key is hold down on Windows, it generates several start DTMF events and only one stop when released.

Update zrtp4j-light that implements new features to negotiate ZRP protocol
versions, support the B256 SAS authentication mode, some smaller updates and
bug fixes.