Previously, if `loadImpulseResponse` was called before `prepareToPlay`,
the IR wasn't guaranteed to have loaded before the first call to
`processSamples`.

Now, we flush the queue of pending IR-load commands during
`prepareToPlay`, which should ensure that the most recently-loaded IR is
ready to use immediately.

The startThread/stopThread calls were moved to prevent a thread
sanitizer warning about a race on the vtable of Impl.

* The stereo option now works like in most guitar amplifier simulations : whatever the number of channels in the process function, or defined in the prepare function, the parameter "wantsStereo" allows the class to do all the time mono or stereo processing
* Fixed a few issues when the user tries to change the normalization or trimming option without changing the impulse response itself
* Reduced memory consumption in various places
* Various improvements to the documentation