[ofw] patch: Fix a race in the cl_timer code that caused deadlocks in opensm

[Hefty, Sean]

> 1) Do we want to allow simultaneous callbacks? For one timer? For more than
> one timer?

I would think one timer = one callback, multiple timers = multiple callbacks.

We currently have one timer = multiple callbacks, which defeats the check to see if timer_start is being called from a timer callback because we only track the last callback.

> 2) Do we want to allow timer start to fall (because of no resources?)? If
> yes, design is simpler.

I'm not sure that we have a choice.  The underlying Windows calls can always fail.

> 3) Do we want start to be blocking sometimes (as it is today).

For user space, I don't know that this matters.

> 4) Do we want to allow start in the callback? Stop in the callback?

Start makes sense.  I don't see the point of a user needing to call stop in the callback.