[Update] FaceFusion Speed up

[wangzijian1010]

This pull request includes several changes to improve the functionality and performance of the lite module, particularly focusing on face recognition and restoration. The most important changes include adjustments to the build configuration, updates to file paths, enhancements to the face detection pipeline, and optimizations for multi-threading and performance measurement.

Build Configuration Changes:

• Disabled ENABLE_TEST and ENABLE_TENSORRT options in CMakeLists.txt to streamline the build process.

File Path Updates:

• Updated file paths in examples/lite/cv/test_lite_facefusion_pipeline.cpp to use absolute paths for consistency and reliability. [1] [2]

Face Detection and Recognition Enhancements:

• Added a new function test_tensorrt in examples/lite/cv/test_lite_facefusion_pipeline.cpp to support TensorRT engine for face detection.
• Introduced multi-threading optimizations in lite/trt/cv/trt_face_68landmarks_mt.cpp by resizing vectors and modifying function signatures to include thread IDs. [1] [2] [3] [4] [5] [6] [7] [8]

Performance Measurement:

• Added performance measurement code to various functions in lite/trt/cv/trt_face_restoration.cpp and lite/trt/cv/trt_face_swap.cpp to log preprocessing, inference, and postprocessing times. [1] [2] [3] [4] [5]

Miscellaneous:

• Changed the author name in the header comment of lite/ort/cv/yolofacev8.cpp.
• Added comments for clarity in lite/ort/cv/yolofacev8.cpp.
• Included additional header files for postprocessing in lite/trt/cv/trt_face_recognizer.h and lite/trt/cv/trt_face_restoration.h. [1] [2]

[DefTruth]

LGTM