Fix full-attention layer id mapping for Hybrid models (e.g. Qwen3-Next/Qwen3.5)

[wanzhenchn]

Motivation

Hybrid models (e.g. Qwen3-Next/Qwen3.5): KV pool only exists for full-attention layers, not layer 0. Pick any mapped full-attention layer id to get the v_head_dim

Modifications

Accuracy Tests

Benchmarking and Profiling

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.