[Codegen][CUDA] Fix codegen of cast among vector bfloat16, fp8 and fp4#17741

Hzfengsy merged 1 commit intoapache:mainapache/tvm:mainfrom

MasterJH5574:tvm-dev/2025-03-11-bf16-fp8-vector-castMasterJH5574/tvm:tvm-dev/2025-03-11-bf16-fp8-vector-castCopy head branch name to clipboard