Purpose
Create a tool placeholder name shark_pup which can takes either the model or its runlog (assuming there is standard logging method) and generates a execution graph which will provide enough information on what nodes of the graph (operators?) are called/used. This will help end users identify which sections of their models have been optimized out of the box and which ones need advanced tuning support.
Expected behavior
run shark_pup --input <directory with model | runlog> --output <output_directory>
Expected output
