🐛 [Bug] Input bindings change b/w fast and global partitioner

Bug Description

Global partitioner:

Torch-TensorRT TensorRT Engine:
  Name: fused_0_engine
  Inputs: [
    id: 0
      name: arg0
      shape: [1, 3, 224, 224]
      dtype: Float
    id: 1
      name: _param_constant0
      shape: [16, 3, 3, 3]
      dtype: Float
    id: 2
      name: _param_constant1
      shape: [16]
      dtype: Float
  ]

Fast partitioner:

Torch-TensorRT TensorRT Engine:
  Name: fused_0_engine
  Inputs: [
    id: 0
      name: arg0
      shape: [1, 3, 224, 224]
      dtype: Float
  ]

To Reproduce

Steps to reproduce the behavior:

Expected behavior

Environment

Build information about Torch-TensorRT can be found by turning on debug messages

• Torch-TensorRT Version (e.g. 1.0.0):
• PyTorch Version (e.g. 1.0):
• CPU Architecture:
• OS (e.g., Linux):
• How you installed PyTorch (conda, pip, libtorch, source):
• Build command you used (if compiling from source):
• Are you using local sources or building from archives:
• Python version:
• CUDA version:
• GPU models and configuration:
• Any other relevant information:

Additional context

Start by comparing the fast and global partitioner paths that produce the shown TensorRT engine input listings. Reproduce the differing bindings with a minimal case and verify that both partitioners preserve the expected inputs; the issue provides no named files or tests to begin from.