fix(agent): enable reasoning_content for Z.AI/GLM models

[vominh1919]

Fixes #16533

Problem

Z.AI/GLM models (glm-5.1, glm-4.7, etc.) never return reasoning_content — even on questions that clearly trigger chain-of-thought when the same model is hit directly via cURL.

Root Cause

Z.AI's API uses the thinking parameter (same format as Kimi) to enable reasoning:

extra_body["thinking"] = {"type": "enabled"}

But Hermes only injected the OpenRouter-style reasoning extra_body, which Z.AI silently ignores. There was no Z.AI-specific handling in the reasoning pipeline.

Fix

Three changes across 2 files:

1. run_agent.py — Add _is_zai URL detection for z.ai and open.bigmodel.cn hosts, pass is_zai flag to transport
2. run_agent.py — Add "z-ai/" to OpenRouter reasoning_model_prefixes so Z.AI models get reasoning via OpenRouter
3. chat_completions.py — Add Z.AI thinking extra_body injection, mirroring the existing Kimi pattern

Files Changed

• run_agent.py — 8 lines added (Z.AI detection + OpenRouter prefix)
• agent/transports/chat_completions.py — 11 lines added (thinking extra_body)

[alt-glitch]

Likely duplicate of #11494 — both add preserved thinking support for Z.AI/GLM models via extra_body.thinking parameter injection.

[teknium1]

Thanks for the focused Z.AI investigation. Current main has already superseded the direct-endpoint portion through the provider-profile path, but the OpenRouter portion remains a narrow candidate for salvage.

Problems

• The changed direct Z.AI branch is no longer the live registered-provider path: agent/chat_completion_helpers.py:849-888 resolves zai to a provider profile, and plugins/model-providers/zai/__init__.py:94-108 now emits extra_body.thinking.
• run_agent.py:5362-5373 still omits z-ai/ from the OpenRouter reasoning gate. Since plugins/model-providers/openrouter/__init__.py:122-160 emits reasoning only when that gate is true, this PR's OpenRouter objective is not present on current main.
• Please add coverage through the current OpenRouter profile/transport path; existing direct-Z.AI coverage is at tests/plugins/model_providers/test_zai_profile.py:209-249.

Suggested changes

• Retain only the OpenRouter prefix change in the current gate and test the resulting extra_body.reasoning payload for z-ai/glm-*.
• Do not restore the legacy is_zai transport branch; direct Z.AI behavior belongs in ZaiProfile.

Automated hermes-sweeper review.

[teknium1]

Current main no longer applies registered-provider behavior through this legacy branch: agent/chat_completion_helpers.py:849-888 resolves zai to ZaiProfile, which owns the direct Z.AI wire payload at plugins/model-providers/zai/__init__.py:94-108. Please keep direct Z.AI behavior in that profile rather than add this transport flag.