[Whisper] handle deprecation of forced_decoder_ids (#38232)

* fix

* working saved forced_decoder_ids

* docstring

* add deprecation message

* exception message ordering

* circular import comment

docs/source/en/model_doc/whisper.md

@@ -95,7 +95,7 @@ transcription[0]
 
 ## Notes
 
-- Whisper relies on [`~GenerationMixin.generate`] for inference.
+- Whisper relies a custom [`generate`] for inference, make sure to check the docs below.
 - The [`WhisperProcessor`] can be used for preparing audio and decoding predicted ids back into text.
 
 ## WhisperConfig