Recognize include directories in package file contents

[nightlark]

One of the main sources of false positives/ambiguity comes from packages that vendor a copy of a header file.

Adding an ability to recognize include standard/sysroot include directories in package contents will help significantly, though to recognize some very commonly used libraries like libstdc++ there are some multiarch and common compiler subfolders that need recognizing:

• Directly under usr/include, maybe with one subfolder being omitted as allowed? (e.g. if they included a vendored copy but left out a subfolder from the include path)
• Multiarch triplet include folders: usr/x86_64-linux-gnu/include (Support for multiarch include paths: https://wiki.debian.org/Multiarch)
• Nested subfolders specific to libstdc++/compilers: usr/x86_64-linux-gnu/include/c++/11/, usr/lib/llvm-16/lib/clang/16/include/stdarg.h, usr/lib/gcc-cross/x86_64-linux-gnux32/12/include/stdbool.h, etc
• Encountered many matches for anything in libstdc++, libc++, and libc across versions (e.g. bits/ and c++/9/ and c++/10/) and architectures need to be collapsed into one (include file name is not enough to determine version): "universe/devel/libstdc++-10-dev-arm64-cross", "usr/aarch64-linux-gnu/include/c++/10/valarray", etc

Examples:

• r-cran packages
• ASIO... Include: asio.hpp -> Packages: [("universe/devel/libasio-dev", "usr/include/asio.hpp"), ("libdevel/libboost1.83-dev,universe/libdevel/libboost1.74-dev", "usr/include/boost/asio.hpp"), ("universe/libdevel/libwebsocketpp-dev", "usr/include/websocketpp/common/asio.hpp"), ("universe/libdevel/libwebsocketpp-dev", "usr/include/websocketpp/config/asio.hpp")]

[nightlark]

1/23 - Steven: has something that works using regex, working on refactoring to do regex on individual path components

• could reuse the individual path component regex code to implement full path normalization and path comparison functions