Determine how different sorts of file names should be normalized

[nightlark]

Different file names need to be normalized to have good odds of finding a match in our datasets. Generally, these are centered around a few things:

• Removing identifiers for a specific version from files names
• Removing platform/architecture specific information from file/folder names

What needs to get done may vary based on the type of file:

• Linux shared libraries (file name needs normalizing to remove architecture and version identifiers)
• C/C++ headers, pkgconfig/CMake Config files (need to recognize common include file paths, including ones that specify a multiarch triplet or GNU triplet)
• Linux binaries (need to recognize common bin folder paths, including ones that specif a multiarch triplet and maybe GNU triplet)
• Language/Ecosystem specific - different languages or ecosystems like Python, Windows, NuGet, and macOS could require handling the above in ways that are different than Linux