- Implemented InterlockedBitTestAndReset, InterlockedBitTestAndSet, InterlockedExchan...

- Rundown re-implementation:

* Added inlined functions for internal system use for quickest path.
* Correctly named all functions Exf instead of Ex.
* Removed PAGED_CODE(); macro where it shouldn't be used.
* Addded multiple ASSERTS for sanity checks.
* Used macros for win64/32 portability.
* Fixed the following bugs/features:
    * ExfAcquireRundownProtection:
        ** Added specific code instead of calling the generic function. Rundown locks are
           performance critical and a dedicated path is prefered.
    * ExfAcquireRundownProtectionEx:
        ** Added a quick immediate check to see if the rundown is active.
    * ExfReleaseRundownProtection:
        ** Added specific code instead of calling the generic function. Rundown locks are
           performance critical and a dedicated path is prefered.
    * ExfReleaseRundownProtectionEx:
        ** Simplified the loop code.
        ** Fixed a bug in signaling of the event during waitblock count removal
    * ExfWaitForRundownProtectionRelease:
        ** Add quick case when we don't actually need a full wait.
        ** Simplified loop code.
* Added stubs for cache-aware implementation.
* Documented the functions.

svn path=/trunk/; revision=20435