model: Remove uses of custom linear op

[rebel-joonsoo]

Pull Request Description

Type of Change

• New Model Support
• 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)
• Documentation update
• Performance improvement
• Code refactoring
• Other (please describe):

Changes Overview

The uses of custom linear ops are removed.

Motivation and Context

The custom linear op was introduced because it takes very long time to trace when tracing with torch.jit.trace. But now the models are now traced by torch export, so the custom linear op is not necessary anymore.

Checklist

• I have performed a self-review of my own code
• I have added tests that prove my fix is effective or that my feature works (If needed)

Additional Information

Related Issues

---

Conventional commit

type(optional scope): description

---

Type candidate

• Model Updates
  • model: Adding New models or Bugfix for existing models
    • ex) Add LlavaNext
    • ex) Bugfix Whisper
• Enhancements
  • performance: Optimizing some models or this library itself
    • ex) Loading RBLNModel faster
    • ex) Optimizing Memory Usage of DecoderOnlyModel
• Code Refactor
  • refactor: Re-arrange class architecture, or more.
    • ex) Refactor Seq2Seq
• Documentation
  • doc: Update docstring only
• Library Dependencies
  • dependency: Update requirements, something like that.
• Other
  • other: None of above.
    • ex) ci update
    • ex) pdm update