Codebase Refactor: Localize wildcard imports

In general, it's best practice to reduce wildcard imports wherever possible so that it's clear where methods/functions are coming from.

Using ruff's F405 rule, we can see that the following files contain wildcard imports where there are names that are undefined without considering wildcard imports. By specifying what comes from where, we can easily figure out what's important:

retired_benchmarks/translation/gnmt/tensorflow/loadgen_gnmt.py
retired_benchmarks/vision/medical_imaging/3d-unet-brats19/Task043_BraTS_2019.py
tools/submission/submission_checker/checks/accuracy_check.py
tools/submission/submission_checker/checks/compliance_check.py
tools/submission/submission_checker/checks/measurements_checks.py
tools/submission/submission_checker/checks/performance_check.py
tools/submission/submission_checker/checks/power_check.py
tools/submission/submission_checker/checks/system_check.py
tools/submission/submission_checker/loader.py
tools/submission/submission_checker/results.py
tools/submission/submission_checker/utils.py
vision/classification_and_detection/python/backend_ncnn.py
vision/medical_imaging/3d-unet-kits19/base_SUT.py
vision/medical_imaging/3d-unet-kits19/inference_utils.py
vision/medical_imaging/3d-unet-kits19/preprocess.py
vision/medical_imaging/3d-unet-kits19/unet_pytorch_to_onnx.py

Start by running Ruff's F405 check against the listed files, then inspect each wildcard import and the names it provides. The affected entry points include the submission checker modules under tools/submission/submission_checker/, the retired benchmark files, and the vision benchmark files. Done means the imports are localized in all listed files and F405 no longer reports these cases.