Add popular model IO configs

[xiaoyu-work]

Describe your changes

1. Add popular model IO configs.
2. Use Olive built in IO configs for ONNX conversion pass.
3. Remove optimum dependency.

This change is verified by running capture-onnx-graph CLI for 70+ huggingface models.

Checklist before requesting a review

• Add unit tests for this change.
• Make sure all tests can pass.
• Update documents if necessary.
• Lint and apply fixes to your code by running lintrunner -a
• Is this a user-facing change? If yes, give a description of this change to be included in the release notes.

(Optional) Issue link

[github-advanced-security]

lintrunner found more than 20 potential problems in the proposed changes. Check the Files changed tab for more details.

[Copilot]

Pull request overview

This PR adds a comprehensive internal implementation of HuggingFace model IO configurations to replace the dependency on the optimum library. The changes include adding support for popular model types and diffusers pipeline components (SD, SDXL, SD3, Flux, Sana), implementing ONNX export configurations, and updating the DiffusersModelVariant enum from "sd15" to "sd".

Changes:

• Adds a new olive/common/hf/io_config/ module with base classes, model configs, input generators, normalized configs, and task definitions for ONNX export
• Updates DiffusersModelVariant enum: renames SD15 to SD, adds SD3 and SANA variants, and moves to olive/constants.py
• Adds DiffusersComponent enum for pipeline component names
• Refactors diffusers model detection to use config files instead of model name patterns
• Updates ONNX conversion to use new io_config instead of optimum's main_export
• Updates all tests and CLI to use the new variant names

Reviewed changes

Copilot reviewed 27 out of 29 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
olive/constants.py	Adds DiffusersModelVariant and DiffusersComponent enums
olive/common/hf/io_config/	New module with 1000+ lines implementing ONNX export configs for 80+ model types
olive/model/handler/diffusers.py	Updates variant detection logic and adds exportable component methods
olive/passes/onnx/conversion.py	Refactors diffusers export to use new io_config instead of optimum
olive/common/hf/model_io.py	Simplifies to use new io_config module
test/ files	Updates all tests to use new variant names and adds comprehensive test coverage
olive/cli/	Updates CLI to support new variants

[shaahji]

Rewrite

return model_type in _ONNX_CONFIG_REGISTRY and task in _ONNX_CONFIG_REGISTRY[model_type]

[xiaoyu-work]

Done!

[shaahji]

Worth converting this if/else to a map, may be even a constant map?

[xiaoyu-work]

Using map now.