Portable C binding

[squell]

This will close #818 (verified on i386).

Note that the bindgen self-test that I edited don't really add much value: they're essentially testing that the [repr(C)] layout is as we expect it to be, but they go a bit beyond what a C compiler would have to guarantee; and even if the layout of these structs is different, it would still mean our PAM bindings would be correct anyway. The only important bit is that the Rust spec given matches the actual C prototypes.

Which is a given, since PAM is an extremely stable API.

This can be reviewed commit-by-commit.

[squell]

Note: we should probably also add "build on a 32-bit platform" to the CI after this pull request.

Also, perhaps the bindgen scripts might also be removed (or move to a "tools" directory for future reference), since running bindgen shouldn't be needed any more.

[codecov]

Codecov Report

Attention: 15 lines in your changes are missing coverage. Please review.

Comparison is base (2fa9234) 57.36% compared to head (3260252) 57.78%.
Report is 9 commits behind head on main.

Files	Patch %	Lines
src/pam/mod.rs	0.00%	8 Missing ⚠️
src/pam/sys.rs	89.70%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #822      +/-   ##
==========================================
+ Coverage   57.36%   57.78%   +0.41%     
==========================================
  Files          74       74              
  Lines       10559    10590      +31     
==========================================
+ Hits         6057     6119      +62     
+ Misses       4502     4471      -31     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.