Issue with c_bool in MpiLock

[mathomp4]

In testing pFlogger v1.16.0, it was found the use of c_bool in MpiLock.F90 was causing issues where MAPL tests using more than 2 processes would lock. So v1.16.1 reverted this:

v1.16.0...v1.16.1#diff-e3c711f0e13147ba4d455afdf37d435ffaa4bb8dc1450b95eee54b9152ba3cdd

However, @tclune noted that using it (compared to a logical) would probably save memory. But something else must be out of sync with using c_bool.