LLM: Optimize qwen1.5 moe model

[hzjane]

Description

• [x] convert moe block
• [x] RMSNorm llama_rms_norm_forward
• [x] MLP llama_mlp_forward
• [x] kv_cache transformers-v4.40.4 past_key_value.seen_tokens -> past_key_value._seen_tokens
• [x] fused qkv
• [x] fused rope
• [x] flash_attention(first_token)
• [ ] esimd_sdp(next_token)
• [ ] quantize kv cache

1. Why the change?

2. User API changes

3. Summary of the change

4. How to test?

• [ ] N/A
• [ ] Unit test
• [ ] Application test
• [ ] Document test
• [ ] ...

5. New dependencies

• [ ] New Python dependencies - Dependency1 - Dependency2 - ...
• [ ] New Java/Scala dependencies and their license - Dependency1 and license1 - Dependency2 and license2 - ...