feat(renderers): register Qwen3.8 Flash Next

[S1ro1]

Summary

• route Qwen/Qwen3.8-Flash-Next through the native Qwen3.8 renderer
• keep thinking enabled by default for the model

Validation

• uv run pytest tests/test_qwen38.py -q (10 passed)
• git diff --check

Note

Register Qwen/Qwen3.8-Flash-Next in MODEL_RENDERER_MAP and qwen3.8 defaults

• Maps Qwen/Qwen3.8-Flash-Next to the qwen3.8 renderer in base.py so auto-selection routes it correctly
• Sets enable_thinking to True by default for Qwen/Qwen3.8-Flash-Next in qwen35.py when unset in config
• Behavioral Change: requests to Qwen/Qwen3.8-Flash-Next now open a `` block by default unless the caller disables it

^{Macroscope summarized d87e8e4.}

[hallerite]

already added in #141