Render-side pre-processing in APM.

This CL adds a way to insert a custom render-side pre-processor to
APM. The pre-processor operates in full-band mode before anything
else. Currently the render processing chain is (if everything is
enabled):

Network --> [Pre processing] --> [Band split] -->
[IntelligibilityEnhancer] --> [Echo canceller (read-only)] -->
[Band merge] --> Playout

Since the render pre processor and capture post processor have the
same interface, I renamed webrtc::PostProcessing into
webrtc::CustomProcessing.

The old APM factory method PostProcessing will be deprecated and
dependencies updated as part of webrtc:8665

NOTRY=True

Bug: webrtc:8665
Change-Id: Ia381cbf12e336d6587406a14d77243d931f69a31
Reviewed-on: https://webrtc-review.googlesource.com/29201
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21327}

modules/audio_processing/test/audio_processing_simulator.cc

@@ -348,7 +348,7 @@ void AudioProcessingSimulator::CreateAudioProcessor() {
     apm_config.residual_echo_detector.enabled = *settings_.use_ed;
   }
 
-  ap_.reset(AudioProcessing::Create(config, nullptr,
+  ap_.reset(AudioProcessing::Create(config, nullptr, nullptr,
                                     std::move(echo_control_factory), nullptr));
   RTC_CHECK(ap_);
 

modules/audio_processing/test/debug_dump_test.cc

@@ -142,6 +142,7 @@ DebugDumpGenerator::DebugDumpGenerator(const std::string& input_file_name,
       apm_(AudioProcessing::Create(
           config,
           nullptr,
+          nullptr,
           (enable_aec3 ? std::unique_ptr<EchoControlFactory>(
                              new EchoCanceller3Factory())
                        : nullptr),