GLM-ASR Support (#42875)

* draft only

* update

* rename

* format config

* update config

* update shape

* update

* testing

* new config

* Update modeling_glmasr.py

* draft change

* update encoder config

* darft PR with no embed

* revert

* revert

* Update modular_glmasr.py

* update

* format

* for doc update

* add processor

* rename

* update rotray

* update convert logtic

* update

* update

* update

* update

* update

* use //2

* update for pytest

* change modular

* for sglang support

* Update modeling_glmasr.py

* audioflamingo3: use _get_audio_token_length

* clean modular

* integration tests

* Glmasr -> GlmAsr

* audioflamingo3: audio len and default prompt as attributes

* modular update

* fix

* udpate integration tests

* use auto dtype

* conversion script update

* fix

* correct test values

* make

* test update

* fix

* make

* auto updates

* doc update

* update doc

* fix

* fix import

* doc update

* fix

* fix

* doc update

* make style

* update max_audio_len

* update repo

* make

* make

* make

---------

Co-authored-by: eustlb <94853470+eustlb@users.noreply.github.com>
Co-authored-by: Eustache Le Bihan <eulebihan@gmail.com>