Added TorchAO FP8 weight-only and static quantization + user guides

[micwill755]

Description

Adds TorchAO FP8 support to the Dynamo path: weight-only (dequantize_affine → TensorRT IDequantizeLayer) and static FP8 (non-decomposed quantize_affine_float8 / dequantize_affine_float8). Constant folding is updated so those DQ ops stay in the graph and are not folded into dense BF16/FP16 weights.

This matches the existing ModelOpt INT8/FP8/FP4 QDQ flow, but for TorchAO tensor-subclass checkpoints. Examples cover a toy nn.Linear (WOQ and static) and FLUX.1-dev WOQ. User-guide, troubleshooting, and gallery docs are updated. HTML under docs/ is left to the post-merge docgen bot.

Dependencies: torchao (examples list it in examples/dynamo/requirements.txt). FP8 compile/runtime needs Hopper or newer (compute capability ≥ 9).

Fixes # (issue)

Type of change

Please delete options that are not relevant and/or add your own.

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

Checklist:

• My code follows the style guidelines of this project (You can use the linters)
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas and hacks
• I have made corresponding changes to the documentation
• I have added tests to verify my fix or my feature
• New and existing unit tests pass locally with my changes
• I have added the relevant labels to my PR in so that relevant reviewers are notified

[micwill755]

Closing this fork-headed PR so we can reopen the same commits from pytorch/TensorRT branches. GitHub native Stacks (1/2, 2/2) cannot include PRs from forks.

[micwill755]

Replacement (same-repo stack 1/2): #4589