Add FM synthesis with 2 operators

[jpcima]

FM example

<control>
set_cc21=100

<global>
volume=-3.0

<region>
sample=*sine
oscillator=on
oscillator_mode=2
oscillator_multi=2
oscillator_detune=5
oscillator_mod_depth=100
oscillator_mod_depth_oncc21=1000

[paulfd]

I think it's filled at the beginning of renderBlock?

[jpcima]

Voices are rendered into a temporary buffer. This could go, however this leads to another problem:
Synth.cpp tempSpan->fill(0.0f); -> called once per cycle
The tempSpan will have to be cleared each time before the voice rendering.