* Support evaluating languages that do not yet have a Bergamot model available
Fixes#106
* Move cached results to top level evaluation directory
* Add Indonesian
* Find right NLLB lang code dynamically
Fixes#103
* Decrease partition size for Google translate
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Bosnian
* Add Danish
* Add Swedish
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Galician
* Add Hindi
* Add Croatian
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Japanese
* Install dependencies required for Japanese tokenization in sacrebleu
* Install dependencies required for Korean tokenization in sacrebleu
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Romanian
* Retry up to 7 times when Google translate's service is unavailable
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Build a single report with all models and results
* Add Slovak
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Vietnamese
* Update evaluation results [skip ci]
* Add Latvian
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Serbian
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Thai
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Korean
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Add Chinese
* Update evaluation results [skip ci]
* Update model registry [skip ci]
* Move logic to fill Bergamot supported languages in a separate function