Adding more refined control over choice of band-splitting

This CL allows the user to have more refined control over what band
splitting-scheme is used inside the audio processing module.


Bug: webrtc:6181
Change-Id: I236c3b1f96ab80cc4ffb8c39c045c034764567a1
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/152480
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29189}

modules/audio_processing/include/audio_processing.h

@@ -245,6 +245,17 @@ class AudioProcessing : public rtc::RefCountInterface {
   // submodule resets, affecting the audio quality. Use the RuntimeSetting
   // construct for runtime configuration.
   struct Config {
+    // Sets the properties of the audio processing pipeline.
+    struct Pipeline {
+      Pipeline();
+
+      // Maximum allowed processing rate used internally. May only be set to
+      // 32000 or 48000 and any differing values will be treated as 48000. The
+      // default rate is currently selected based on the CPU architecture, but
+      // that logic may change.
+      int maximum_internal_processing_rate;
+    } pipeline;
+
     // Enabled the pre-amplifier. It amplifies the capture signal
     // before any other processing is done.
     struct PreAmplifier {