matroskamux: reduce muxing overhead with audio-only streams
@tpm
Submitted by Tim Müller Link to original bug (#776326)
Description
+++ This bug was initially created as a clone of Bug 754696 +++
We might need to add a property to configure the max interleaving delay for audio-only streams, so we can lace multiple audio frames into one cluster and only mark the cluster start as keyframe for multifdsink, in order to avoid the extra mux overhead of putting each audio frame into its own cluster.