gst/audioresample/: Fix audioresample, seek torture, new segments, reverse negotiation etc.. work fine.

Original commit message from CVS:
* gst/audioresample/buffer.c: (audioresample_buffer_queue_flush):
* gst/audioresample/buffer.h:
* gst/audioresample/gstaudioresample.c:
* gst/audioresample/gstaudioresample.h:
* gst/audioresample/resample.c: (resample_input_flush),
(resample_input_pushthrough), (resample_input_eos),
(resample_get_output_size_for_input),
(resample_get_input_size_for_output), (resample_get_output_size),
(resample_get_output_data):
* gst/audioresample/resample.h:
* gst/audioresample/resample_ref.c: (resample_scale_ref):
Fix audioresample, seek torture, new segments, reverse negotiation
etc.. work fine.