[NVPTX] Add support for PTX ISA v8.8 (#136639)
Support PTX version 8.8 (`-mattr=+ptx88`) from CUDA 12.9. The following
new targets are also added:
- SM103 and SM121: sm_103, sm_103a, sm_121, sm_121a.
Also, some things were reformatted.
https://docs.nvidia.com/cuda/parallel-thread-execution/#changes-in-ptx-isa-version-8-8