Been wanting to do this for a long time.
This does impose performance issues when building with the vanilla
release profile, as cargo does not perform optimizations across crate
boundary (we care mostly about inlining).
so I added release-lto profile
Also fixed some broken stuff I found across the project, meh
Former-commit-id: 06d03263cc6245313f3ea22c715479ab6da7c4d4
Former-commit-id: f93abd10c67ea8a3b8072b47462be5eca4f3e02b
