- yield was in wrong if block, not sure how

- count ULTs in progress pool that are waiting on communication as
  waiters so that we don't busy spin when it isn't necessary
- if a trigger occurs, yield immediately to service ULTs that may have
  been activated as a result
- HG_Progress() is still called too frequently, see TODO note
- outcome: this makes servers significantly faster, but seems to make
  clients a bit slower; optimal performance right now comes from
  building servers with this git commit and building clients without
  it; need to investigate

in cases where margo_forward_timed calls timeout, margo will
return HG_TIMEOUT rather than HG_CANCELED

we need to remember whether a sleeping thread is in the same pool
as the progress thread to properly decrement the number of
waiting ULTS.

I factored the sorted queue implementation Phil created for
monitoring timeouts out into separate files (margo-timer.c/h).
There is now an interface for submitting timers that define
a callback function that's executed when the timers expire.
margo_thread_sleep() and margo_forward_timed() will both use
this interface, though only margo_thread_sleep() is implemented
at the moment.

- na_ and NA_ types and functions no longer necessary

- also a new version of utlist.h with APPEND_ELEM macro

- Demonstrated in test program.  This function allows caller to suspend
  until some other entity calls margo_finalize(). If called by a ULT
  that is executing in the same pool as the HG progress thread, then it
  will indicate to the progress thread that it does not need to yield
  execution.
- fixes #6

get_size() is the right function, just keep in mind that it doesn't
count the caller, only work units that are queued for future execution
in the pool

- fixes #5
- suggested by John
- run example server as "./server single" intead of "./server" to
  demonstrate