Hume

- Introduces `VoiceTool` to enable function calling and other tools
- Adds `create_tool()`, `get_tool()`, `iter_tools()`, and `delete_tool()` to the `HumeVoiceClient`
- Adds `LanguageModelConfig` to update the LLM used as well as the temperature parameter
- Adds `VoiceIdentityConfig` to enable updating the voice of EVI
- Adds `send_json` method on the `VoiceSocket` to enable sending tool responses

- Introduces `HumeVoiceClient`, a Python client for interacting with EVI, Hume's new Empathic Voice Interface.
- Introduces a `MicrophoneInterface` that allows you to talk to EVI through your device's microphone and speakers.
- Adds support for creating, getting, listing, and deleting EVI configs with `HumeVoiceClient`.
- Adds support for configuring a `VoiceSocket` with an EVI config through the `HumeVoiceClient`.
- Adds support for configuring the voice of EVI to be either Ito or Kora.
- Adds support for configuring the session and audio settings on a `VoiceSocket` with `update_session_settings()`.
- Adds support for listing chat history with `HumeVoiceClient`.

- Various dependency upgrades

- Adds `text` parameter to `HumeBatchClient.submit_job()` to support passing raw text as the input for batch jobs.

- Supporting Python 3.10 and 3.11
- Dropping support for Python 3.8 (note that the package will still install in 3.8 environments, but compatibility will not be maintained)

- The `raise_on_failed` in `BatchJob` will raise an exception if a `BatchJob` fails
- When using `HumeStreamClient` models can be configured separately for each payload. Configuring models when initializing the `StreamSocket` is still supported.
- Support for `fps_pred`, `prob_threshold`, and `min_face_size` when using `FaceConfig` with `HumeStreamClient`