[Feature Request][PyTorch] Support thd format for fp8 tensors in DotProductAttention

[alexdremov]

Seems like at the current time packed tensors in thd format are not supported by transformer_engine.pytorch.attention. DotProductAttention. That's weird as such mode clearly supported by fused_attn_fwd from fused_attn cpp_extensions

I see that FusedAttnFunc was used in FusedAttention, but implementations for FusedAttnFunc_kvpacked and FusedAttnFunc_qkvpacked are not present. I suppose, they could be added in a same way

[yiakwy-xpu-ml-framework-team]

Thd is going to be supported in TensorEngine. So if you use TensorEngine you will get it.