[WIP] add more example for permute/scatter-gather/moe

[carlushuang]

• [x] add test topk
• [x] add test topk-softmax
• [x] add test tile_reduce
• [x] add test scatter-gather
• [x] add tensor transform support for scatter-gather
• [x] modify buffer raw related tile api
• [x] add async load (non-raw version) api
• [x] add block_tile_reduce_xor_sync() api
• [x] add example permute
• [ ] add permute utility kernel for moe index
• [ ] debug moe-ffn pipeline
• [ ] add example moe