Support Fusion for ReadPutPacket Operation at DSL (#742)

Support is being added for fusing the ReadPutPacket operation on DSL,
which reduces the overhead caused by reading packet data multiple times
in the scratch buffer. Fusion will occur when two rppkt operations are
executed consecutively with the same src_buffer:

rppkt(src, dst0) + rppkt(src, dst1) -> rppkt(src, [dst0, dst1]

Co-authored-by: Binyang Li <binyli@microsoft.com>

python/test/executor_test.py

@@ -11,7 +11,7 @@ from mscclpp import (
     env,
 )
 from mscclpp import CommGroup, GpuBuffer
-from mscclpp.utils import KernelBuilder, GpuBuffer, pack
+from mscclpp.utils import KernelBuilder, pack
 import os
 import struct