Export mscclpp GpuBuffer to dlpack format

[Binyang2014]

For mscclpp, to use nvls we require the buffer is allocated by mscclpp::GpuBuffer. Due to cupy doesn't support bfloat16 yet, we export the raw buffer to dlpack format.
User can use this feature to create buffer with type supported by pytorch

buffer = RawGpuBuffer(1024 * 2) # 2 for bfloat16
dl_pack = buffer.to_dlpack(str(torch.bfloat16))
tensor = torch.utils.dlpack.from_dlpack(dl_pack)

[chhwang]

Looks good to me, but we would need to support multi-dimensional allocation directly. If users convert 1D to multi-dim, it may incur a local copy which is not intended here.

[chhwang]

LGTM