Fix issue 2115450: a2dp thread is started, even though we are only connected to headset and not playing music.

This is due to a regression introduced by change 24114: when no audio tracks are ready for mixing, 0s are written to audio hardware. However this should only happen if tracks have already been mixed since the audio flinger thread woke up.
Also do not write 0s to audio hardware in direct output threads when audio format is not linear PCM.
diff --git a/libs/audioflinger/AudioFlinger.cpp b/libs/audioflinger/AudioFlinger.cpp
index 790a655..d9be007 100644
--- a/libs/audioflinger/AudioFlinger.cpp
+++ b/libs/audioflinger/AudioFlinger.cpp
@@ -1239,7 +1239,7 @@
                 // active tracks were late. Sleep a little bit to give
                 // them another chance. If we're too late, write 0s to audio
                 // hardware to avoid underrun.
-                if (sleepTime < kMaxBufferRecoveryInUsecs) {
+                if (mBytesWritten == 0 || sleepTime < kMaxBufferRecoveryInUsecs) {
                     usleep(kBufferRecoveryInUsecs);
                 } else {
                     memset (curBuf, 0, mixBufferSize);
@@ -1741,7 +1741,8 @@
                 standbyTime = systemTime() + kStandbyTimeInNsecs;
             } else {
                 sleepTime += kBufferRecoveryInUsecs;
-                if (sleepTime < kMaxBufferRecoveryInUsecs) {
+                if (mBytesWritten == 0 || !AudioSystem::isLinearPCM(mFormat) ||
+                    sleepTime < kMaxBufferRecoveryInUsecs) {
                     usleep(kBufferRecoveryInUsecs);
                 } else {
                     memset (mMixBuffer, 0, mFrameCount * mFrameSize);