Add OpenVoice VC models

[eginhard]

The OpenVoice v1 and v2 voice conversion models can now be used through Coqui:

from TTS.api import TTS

ov1 = TTS("voice_conversion_models/multilingual/multi-dataset/openvoice_v1")
ov2 = TTS("voice_conversion_models/multilingual/multi-dataset/openvoice_v2")

src = "source.wav"
ref = "target.wav"

ov1.voice_conversion_to_file(src, ref, "ov1.wav")
ov2.voice_conversion_to_file(src, ref, "ov2.wav")

And via CLI:

tts --model_name voice_conversion_models/multilingual/multi-dataset/openvoice_v1 \
    --out_path ov1.wav --source_wav source.wav --target_wav target.wav
tts --model_name voice_conversion_models/multilingual/multi-dataset/openvoice_v2 \
    --out_path ov2.wav --source_wav source.wav --target_wav target.wav