Add elfIsPie to check for position-independent executables

[RyanGlScott]

This adds elfIsPie, a function which returns True if the supplied ElfHeaderInfo corresponds to a position-independent executable (PIE). The logic for elfIsPie is adapted from the source code for the file binutils program, which was in turn summarized here: https://unix.stackexchange.com/a/435038

This adds a directory dependency to elf-edit, as directory is used to check file permissions as a fallback. This dependency should be cheap, as directory is a GHC boot library.

Fixes #46.

[langston-barrett]

Love it!

[kquick]

Minor quibble: PIE is an acronym (like ELF), so I would have preferred the name elfIsPIE. I know we aren't consistent on this, just quibbling.

It would be good to document the "is the file executable" fallback in the haddocks themselves. I'm not sure what percentage of the time the fallback is needed, but it would probably be surprising for people to know that file metadata and not ELF data can control the PIE result. I can easily see a scenario where an ELF file is copied for analysis without maintaining executable privileges...

[RyanGlScott]

PIE is an acronym (like ELF), so I would have preferred the name elfIsPIE.

Right... I am increasingly finding myself on the side of not capitalizing all of the letters in acronyms when converting them to lowerCamelCase Haskell names, as it leads to strangeness like dynPLTRel. (Is the acronym PLT or PLTR?) That being said, I don't feel strongly about this—I can change this if you want.

It would be good to document the "is the file executable" fallback in the haddocks themselves.

Sure thing. I've opened #48 for this.

[kquick]

But is dynPltRel checking dynamically that P is less-than Rel? :-) I don't feel strongly enough to cause PR churn though.