Normalize package names

[nightlark]

After looking at normalizing Linux shared library (.so) file names in #5 and looking at file names/package names for a bunch of stuff in Ubuntu package, there were a number of cases where package names contain a version number or dev suffix: #5 (comment)

Some functions should be written to extract the version number from the package names in these cases and normalize them in some way to a name that doesn't contain version identifiers. This could involve mapping them to their source package name, though there are corner cases such as llvm/gcc which contain "subpackages" that would be better described as "libc", "libstdc++", or "clang" rather than the source package name (which might not even be llvm: it could be llvm-defaults or llvm-runtime).