Fuse rmsnorm and quant

[apinge]

Motivation

This PR fuses dynamic quant kernel into rmsnorm.

Modifications

Accuracy Tests

python3 /opt/sglang/benchmark/gsm8k/bench_sglang.py    --port 9080       --tokenizer-path /model/offical_qwen3p5_397B_ptpc/    --max-new-tokens 4096 
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [01:18<00:00,  2.55it/s]
Accuracy: 0.970
Invalid: 0.000
Latency: 78.430 s
Output throughput: 465.588 token/s

Benchmarking and Profiling

Test with concurrency=10 397B

Before

After

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review Process

1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
4. After green CI and required approvals, ask Merge Oncalls to merge.