* add expand_v2/expand_as_v2 for kunlun

* update expand_as_v2

* update expand_as_v2

* support float16/bool

* update xpu.cmake