[TRT EP] Fix trt_load_user_initializer for large models where weight are not correctly excluded

[gedoensmax]

Description

This change respects initializers that are external but already loaded in memory. This is required due to an optimization that leaves it to the backend to read a mapped memory area.

@chilo-ms can you help run the CI and merge this change ?

[gedoensmax]

Can we get this into 1.23 since it is a fix on top of #25409.

[Copilot]

Pull Request Overview

This PR fixes an issue in the TensorRT execution provider where external initializers (weights) that are already loaded in memory were not being handled correctly for large models. The fix ensures that initializers with external data in memory are properly recognized and processed alongside those with raw data.

• Converts TensorrtUserWeights from struct to class with proper accessors to improve encapsulation
• Adds handling for external initializers that are already loaded in memory using HasExternalDataInMemory checks
• Updates serialization logic to exclude external data from graph proto when include_initializer_data is false

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
tensorrt_execution_provider.h	Converts TensorrtUserWeights to class with proper encapsulation and accessor methods
tensorrt_execution_provider.cc	Adds external data handling logic and updates weight processing to use new class interface
graph_proto_serializer.cc	Updates serialization condition to handle external data in memory alongside raw data

[jywu-msft]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[jywu-msft]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[jywu-msft]

there are build errors (warnings as errors)
/onnxruntime_src/onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.h:173:3: error: type qualifiers ignored on function return type [-Werror=ignored-qualifiers]
173 | const int64_t Size() const {
| ^~~~~
cc1plus: all warnings being treated as errors

[gedoensmax]

Thanks, should be fixed now. I also fixed some test usage and improved the logging.

[jywu-msft]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[gedoensmax]

The failing CI seems tobe unrelated to my changes but had some network outage or builder crashes.

[snnn]

Hi there! We haven't cut the release branch for this version yet, so I'm removing the release:1.23.0 label for now to keep things tidy. Thanks so much for your contribution! We'll make sure this gets included when the release is prepared. 🤖