Before, it was called from individual encoder (de)init procedures, which
lead to spare threads running (which were actually lost, leaked).
The current fix uses trivial approach, and the downside is that the
processing thread is working always, even when there's no consumer.
Signed-off-by: Andrey Utkin <andrey.krieger.utkin@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>