add macOS say backend support

[DiogoDoreto]

Introduce a new TTS backend using the built-in macOS 'say' command for text-to-speech generation. The default backend now automatically selects 'say' on Darwin systems and 'kokoro' elsewhere. This provides native, zero-dependency TTS backend for macOS users.

---

This is a basic implementation based on the documentation found here: https://ss64.com/mac/say.html

I'll leave this PR open until someone can confirm this works, as I don't have this OS available to me.

[DiogoDoreto]

• should probably pair with afplay as a new frontend

[riscy]

I tested this out and it seems to not work on my mac -

Despite what the man page says, "say" does not seem to support wav outputs (at least on my machine) with the default voice (the log shows: [tts-say] Opening output file failed: fmt?, where fmt? is say's obscure way of saying it doesn't work with that file format). It works fine if .aiff is the tts-audio-extension so it may be a good idea to guide users in that direction.

You probably already know this, but macOS "say" also lets you skip the step where you write to a file and invoke ffplay. But ffplay is easy to install on macOS (so I read).