Fix tts_with_vc failing on XTTS when using speaker_wav

[rj25031]

When using XTTS (tts_models/multilingual/multi-dataset/xtts_v2), the helper methods
tts_with_vc() and tts_with_vc_to_file() fail with:

ValueError: Model is multi-speaker but no speaker is provided.

This happens even when speaker_wav is provided, which should be sufficient for voice cloning with XTTS.

After inspecting the source, this appears to be a logic bug in how tts_with_vc internally calls tts_to_file.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}