WaveActiveCountBits.test produces incorrect results for KosmicKrisp on clang only

[farzonl]

The clang results are off here. Which makes it hard to determine if this really is a driver bug.

Expected:
---
Name:            ExpectedOut
Format:          UInt32
Stride:          4
Data:            [ 3, 3, 3, 3, 2, 2, 0 ]
OutputProps:
  Height:          0
  Width:           0
  Depth:           0
...
Got:
---
Name:            Out
Format:          UInt32
Stride:          4
Data:            [ 3, 3, 3, 3, 4, 4, 0 ]
OutputProps:
  Height:          0
  Width:           0
  Depth:           0
...
Full Hex 64bit representation of Expected Buffer Values:
[ 0x3, 0x3, 0x3, 0x3, 0x2, 0x2, 0x0 ]
Full Hex 64bit representation of Actual Buffer Values:
[ 0x3, 0x3, 0x3, 0x3, 0x4, 0x4, 0x0 ]