[souravkr114]

Description
This PR implements the ONNX ThresholdedRelu activation operator in TMVA SOFIE, along with its parser ingestion and upstream unit tests.

Changes Included:

1. Operator Definition:

   • Added ROperator_ThresholdedRelu.hxx in tmva/sofie/inc/TMVA/ supporting scalar fallback emission with branchless ternary formulation:
     tensor_Y[id] = ((tensor_X[id] > alpha) ? tensor_X[id] : 0.0f);

2. Parser Integration:

   • Added ParseThresholdedRelu.cxx in tmva/sofie_parsers/src/ to parse ONNX protobuf nodes and extract the alpha attribute (defaulting to 1.0).
   • Registered ThresholdedRelu in RModelParser_ONNX.cxx and added it to tmva/sofie_parsers/CMakeLists.txt.

3. Tests:

   • Added make_ThresholdedRelu model generator in generate_input_models.py.
   • Added automated GTest TEST(ONNX, ThresholdedRelu) in TestCustomModelsFromONNX.cxx.

Validation:

• Verified end-to-end model ingestion from ONNX and C++ inference code generation locally on Windows (MSVC 19.x).

[github-actions]

Test Results

    22 files      22 suites   3d 10h 28m 22s ⏱️
 3 884 tests  3 869 ✅ 0 💤 15 ❌
76 000 runs  75 972 ✅ 0 💤 28 ❌

For more details on these failures, see this check.

Results for commit 54822dc.