When the number of elements per bucket is too small, then try to reduce
the number of threads by merging buckets. This is to ensure that the
number of running wait queue threads doesn't get too big.

To implement waiting for an arbitrary number of handles, we group them
in buckets up to (MAXIMUM_WAIT_OBJECTS - 1) objects, and then assign a
dedicated wait queue thread. The last handle is used to notify about
changes.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

That can only happen by manually modifying
select_vertex_implementation() and select_fragment_implementation(),
which may be useful for testing non-default vertex / fragment pipeline
implementations.