[PyTorch] Fine-grained recipe docs

[negvet]

Description

Fine-grained recipe documentation.
This is enabled after hybrid quantization support #2817 is merged

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refactoring

Changes

Please list the changes introduced in this PR:

• Change A
• Change B

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[greptile-apps]

Greptile Summary

Adds comprehensive documentation for constructing fine-grained PyTorch quantization recipes and updates the runnable example to accurately describe wgrad input provenance.

• Documents quantizer roles, per-GEMM direction mapping, hybrid quantizer provenance, high-precision directions, readiness levels, and API constraints.
• Adds API references and experimental warnings for hybrid and identity quantization types.
• Provides a hardware-gated training example combining MXFP8, NVFP4, and BF16 representations.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

Important Files Changed

Filename	Overview
docs/features/low_precision_training/fine_grained_quantization/fine_grained_quantization.rst	Adds the main fine-grained quantization guide, including factory contracts, direction mapping, provenance choices, and usage caveats.
docs/features/low_precision_training/fine_grained_quantization/pytorch_fine_grained_quantization_example.py	Adds a runnable custom-recipe example whose updated original-BF16 wgrad description matches the configured quantizers and Linear backward path.
docs/api/pytorch.rst	Adds API entries for quantizer roles and hybrid and identity quantizer tensor types.
transformer_engine/common/recipe/init.py	Clarifies the CustomRecipe factory contract for optional and boundary roles.
transformer_engine/pytorch/tensor/hybrid_tensor.py	Marks HybridQuantizer as experimental without changing runtime behavior.
transformer_engine/pytorch/tensor/identity_tensor.py	Adds experimental-status and constructor documentation without changing runtime behavior.

_{Reviews (2): Last reviewed commit: "Update docs/features/low_precision_train..." | Re-trigger Greptile}