fix: Gemma4-31B CoderForge CP8/64K data processing and recipe (#3446)

* fix(examples): add Gemma4 CoderForge generation-block chat template for assistant masking

Signed-off-by: Abhishree <athittenaman@nvidia.com>

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(datasets): forward processor tokenizer to text datasets in VLM dataloader

Signed-off-by: Abhishree <athittenaman@nvidia.com>

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(recipes): drop dataset.tokenizer in VLM dataloader resolution

Signed-off-by: Abhishree <athittenaman@nvidia.com>

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>