Add batching support to the MLX implementation (#347)

* with batching for stt

* Fixed embeddings bug

* pseudo-working tts

* corrected state bug

* reset cache

* corrected delay bug

* corrected condition attribute multi-speaker bug

* corrected max

* cleaned logs

* cleanup

* Adapted sampling functions

* fixed dimensions

* cleanup and default values

* paralelism for min_p_sampling

* cleanup

* Revert transformer.py to origin/main version

* make flake8 happy

* update version

* Less diff on sampling

* version fix

---------

Co-authored-by: Carmen Bach <[email protected]>

Bump the python package versions. (#341)

Bump the python package versions. (#341)

Bump the version number for moshi. (#334)

Sketch the 0.2.9 release. (#330)

Moshi mlx.0.2.11 (#333)

* Fix the local-web MLX example + port to sphn 0.2.

* Formatting fixes.

* Use sphn 0.2.

* Fix.

* Bump the moshi mlx dependency.

v0.2.8 release

v0.2.7 moshi release

Bump the mlx version. (#325)

Add some hooks to handle streaming. (#322)