fix(agent): route reasoning_content to reasoning channel

[xiwuqi]

📝 Description

Fixes the reasoning channel regression behind #1746: OpenAI-compatible providers can return reasoning in reasoning_content, but PicoClaw did not forward that content to the configured reasoning channel.

Problem summary:

• reasoning_channel_id is expected to receive model reasoning for channels such as Telegram.
• OpenAI-compatible providers can populate reasoning_content while leaving reasoning empty.
• PicoClaw detected that reasoning existed, but the reasoning channel received nothing.

Root cause:

• pkg/agent/loop.go passed only response.Reasoning into handleReasoning().
• For OpenAI-compatible providers, the preserved reasoning often lives in response.ReasoningContent, so the publish path saw an empty string and returned early.

What changed:

• Keep the existing response.Reasoning path unchanged when it is populated.
• Fall back to response.ReasoningContent before calling handleReasoning().
• Add a regression test that reproduces the OpenAI-compatible case and verifies the reasoning message is delivered to the configured channel.

Why this fix:

• It is the smallest change that directly addresses the reported failure without broadening provider normalization behavior.
• It preserves the current priority of response.Reasoning and only uses response.ReasoningContent when needed.

Risk / compatibility:

• Low risk. This only affects the reasoning-channel publish path, and only when response.Reasoning is empty.
• Providers that already populate response.Reasoning keep their existing behavior.

🗣️ Type of Change

• 🐞 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• 📖 Documentation update
• ⚡ Code refactoring (no functional changes, no api changes)

🤖 AI Code Generation

• 🤖 Fully AI-generated (100% AI, 0% Human)
• 🛠️ Mostly AI-generated (AI draft, Human verified/modified)
• 👨‍💻 Mostly Human-written (Human lead, AI assisted or none)

🔗 Related Issue

Fixes #1746

📚 Technical Context (Skip for Docs)

• Reference URL: [BUG] reasoning_channel_id not working with OpenAI-compatible provider in v0.2.3 despite PR #604 #1746
• Reasoning: The agent event path already treats ReasoningContent as evidence that reasoning exists, but the channel delivery path only forwarded response.Reasoning. That mismatch makes OpenAI-compatible reasoning invisible to reasoning_channel_id.

🧪 Test Environment

• Hardware: PC
• OS: Windows (local development environment)
• Model/Provider: N/A (agent/unit tests only)
• Channels: Telegram (simulated via channel manager test double)

📸 Evidence (Optional)

Click to view Logs/Screenshots

Passed:

• go test ./pkg/agent -run TestProcessMessage_PublishesReasoningContentToReasoningChannel -count=1
• go test ./pkg/agent -run TestHandleReasoning -count=1
• go test ./pkg/providers/openai_compat -count=1

Additional broader validation:

• go test ./pkg/agent -count=1 still fails in TestGlobalSkillFileContentChange.
• I reproduced that same failure on a clean upstream/main worktree at commit 3500080, so it is pre-existing and not introduced by this PR.
• codex review --base origin/main could not run in this environment because the local Codex config rejected approvals_reviewer = never.

☑️ Checklist

• My code/docs follow the style of this project.
• I have performed a self-review of my own changes.
• I have updated the documentation accordingly.

[yinwm]

LGTM! This is a well-targeted bug fix with minimal code changes.

Summary:

• Fixes the regression where reasoning_channel_id did not work with OpenAI-compatible providers
• Root cause: handleReasoning only received response.Reasoning, which is empty for OpenAI-compatible providers
• Fix: Fallback to response.ReasoningContent when response.Reasoning is empty

Why approve:

1. Minimal change - 4 lines added, 1 modified
2. Correct priority - keeps Reasoning as primary source
3. Comprehensive regression test added
4. Low risk - only affects reasoning channel publish path
5. Directly addresses issue 1746

Thanks @xiwuqi!