Lock profiling and lock order verification, week 12

[Greg Miller]

Accomplishments for the week:

* Implemented XML output for both lock profiling and libwitness
* Restructured the witness data in order to allow locks with the same
name to be treated as the same lock
* Added additional tests to verify that data for locks with the same
name is properly shared
* Replaced pthread_setname_np() with pthread_mutex_setname_np(),
pthread_rwlock_setname_np(), and pthread_spin_setname_np() functions.
* Remerge from the latest trunk mirror to the branch.

Remaining work:

* Add backtrace logging to libwitness
* Cleanup/documentation