💥 Reduce allocations in VST3 audio sockets

We do this by using this new `MessageReference<T>` type to avoid copying
our `YaAudioProcessor::Process` struct and the contained `YaProcessData`
object. This is only part of the work, but this redesign lets us keep
the these objects alive on both the plugin and the host side. On the
plugin side, we'll simply serialize the data from the referred to object
without copying it. On the Wine side, we'll write the data to a
persistent thread local object, and then reassign the
`MessageReference<T>` to point to that object. This lets us serialize
'references', thus avoiding potentially expensive allocations. With
these last few changes alone VST3 plugins are already at the same
performance level as our optimized VST2 plugin groups.

src/common/serialization/vst3/process-data.cpp

@@ -281,6 +281,9 @@ YaProcessDataResponse YaProcessData::move_outputs_to_response() {
     //       `ProcessData` object generated in `get()` here sicne these of
     //       course are not references or pointers like all other fields, so
     //       they're not implicitly copied like all of our other fields
+    // FIXME: Instead of moving, the `YaProcessDataResponse` should be an
+    //        (optional) field. Moving defeats the point of us trying to reuse
+    //        these objects.
     for (int i = 0; i < reconstructed_process_data.numOutputs; i++) {
         outputs[i].silence_flags =
             reconstructed_process_data.outputs[i].silenceFlags;