[AFV] Change output format and processing to 32 bit float

32 bit float is what is used in C# reference and has a much higher
dynamic range. 16 bit integer was clipping very often with all the VHF
simulation applied.

src/blacksound/sampleprovider/mixingsampleprovider.h

@@ -30,7 +30,7 @@ namespace BlackSound
             void addMixerInput(ISampleProvider *provider) { m_sources.append(provider); }
 
             //! \copydoc ISampleProvider::readSamples
-            virtual int readSamples(QVector<qint16> &samples, qint64 count) override;
+            virtual int readSamples(QVector<float> &samples, qint64 count) override;
 
         private:
             QVector<ISampleProvider *> m_sources;