feat: Add "none" reasoning effort support for GPT-5.2 and GPT-5.1

[ben-vargas]

Summary

Adds support for reasoning_effort: "none" for GPT-5.2 and GPT-5.1 general purpose models, which disables the reasoning phase entirely for faster responses when reasoning is not needed.

Motivation

Per OpenAI API documentation and the Codex CLI implementation:

• GPT-5.1 defaults to none and supports: none, low, medium, high
• GPT-5.2 (being newer than 5.1) also supports none along with low, medium, high, xhigh

References:

• OpenAI API docs - reasoning_effort: "gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high."
• Codex CLI ReasoningEffort enum includes None variant (codex-rs/protocol/src/openai_models.rs)
• Codex CLI request builder passes ReasoningEffort::None through to API without rejection (codex-rs/core/src/client.rs)
• Codex CLI config docs list model_reasoning_effort = "none" as valid (docs/config.md)

Changes

Added

• gpt-5.2-none and gpt-5.1-none model mappings in model-map.ts
• supportsNone detection in getReasoningConfig() for GPT-5.1/5.2 general purpose models
• 4 new unit tests for "none" reasoning behavior (197 total tests)

Changed

• Models that don't support "none" auto-convert to appropriate fallback:
  • Codex / Codex Max → "low"
  • Codex Mini → "medium" (unchanged from before)
• Documentation updated with complete reasoning effort support matrix per model family

Testing

npm run typecheck  # ✅ Pass
npm test           # ✅ 197 tests passing
npm run build      # ✅ Pass

Manually tested with openai/gpt-5.2-none model - confirmed working.

Reasoning Effort Support Matrix

Model	none	low	medium	high	xhigh
GPT-5.2	✅	✅	✅	✅	✅
GPT-5.1	✅	✅	✅	✅	❌
GPT-5.1-Codex	❌→low	✅	✅	✅	❌
GPT-5.1-Codex-Max	❌→low	✅	✅	✅	✅
GPT-5.1-Codex-Mini	❌→medium	❌→medium	✅	✅	❌→high

Notes

• This plugin defaults to "medium" for better coding assistance; users must explicitly configure "none" if desired
• The OpenAI API docs state GPT-5.1 defaults to "none", but we intentionally keep "medium" as our default since this is a coding assistant plugin