Enable MobiusBuilder in olive optimize

[Copilot]

Describe your changes

• Exporter selection

  • Add mobius_builder to olive optimize --exporter.
  • Keep model_builder as the default text exporter.

• Pass scheduling

  • Emit MobiusBuilder with the requested fp32/fp16/bf16 precision.
  • Reject unsupported Mobius quantized precisions.
  • Skip redundant peephole and fp16 conversion passes for Mobius output.

olive optimize -m <model> --exporter mobius_builder --precision fp16

• Coverage and docs
  • Add pass-list coverage for Mobius export.
  • Document the new exporter option.

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.
  • olive optimize now supports --exporter mobius_builder for fp32, fp16, and bf16 ONNX capture.

(Optional) Issue link

• Fixes [FR]: Enable mobius exporter in optimize cli command #2639

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.

[devang-ml]

Mobius builder should support input model that is already quantized in int4