Add FP8*(B)F16 GEMM

[lzhangzz]

• Add FP8*(B)F16 for sm_70 ... sm_90
• Optimize grouped GEMM performance for all mixed GEMMs
• Re-organized code structure

[lvhan028]

In lmdeploy chat, when typing exit, it crashed:

terminate called after throwing an instance of 'std::system_error'
  what():  Resource deadlock avoided
Aborted (core dumped)