jdupes 1.18.2 - Google Drive File Stream fix

For some reason, Google Drive File Stream hands down the same device:inode pairing for all directories, which technically means (to jdupes) that all directories are the same directory. Since this is very obviously not the case, the double-traversal protection code needs to be turned off for the -r/--recurse option to work on Google Drive File Stream. That code exists for a very good reason (it prevents you from accidentally adding the same directory twice and having every file match itself and potentially get deleted at the end, plus it blocks symbolic links to directories from causing an infinite loop of recursion) but it makes it basically impossible to make any practical use of jdupes on the virtual hard drive that GDFS gives you to work with.

This release adds the -U/--notravcheck option which disables the double-traversal protection mechanism, allowing GDFS to work with recursion as expected. There isn't any real reason to worry about updating unless you need this feature.

Also included in this release's code base is a unified "package" build script (just puts binaries and documentation in a zipped or tarred directory) and build-time removal of the CPU cache line info code (that only works on Linux) when not compiling for Linux. I also corrected some documentation typos and split the internal behavior flags since the program ran out of 32-bit flag space. There are too many features to fit in a mere 32 bit flags!