What does this PR do?

Fixes GLM/ZAI vision request compatibility when auxiliary.vision is configured via a custom endpoint (provider: glm + base_url + api_key).

The bug path was:

• provider resolves to custom
• _build_call_kwargs only skipped max_tokens for provider == "zai"
• GLM vision endpoints (bigmodel/z.ai) reject this with 1210 parameter errors

This change extends the same guard to custom endpoints that resolve to bigmodel.cn or z.ai for vision model names, and hardens 1210 retry handling by also dropping temperature on the ZAI-specific retry path.

Related Issue

Fixes #26827

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• Updated agent/auxiliary_client.py:
  • _build_call_kwargs now skips max_tokens for vision models when provider == "custom" and base URL host matches bigmodel.cn or z.ai.
  • Sync/async call_llm retry path now strips temperature as well when handling ZAI 1210 parameter errors.
• Added tests in tests/agent/test_auxiliary_client.py:
  • custom bigmodel/z.ai vision endpoints do not emit max_tokens
  • non-ZAI custom endpoints keep max_tokens

How to Test

1. Run targeted tests:
   • uv run --with pytest==9.0.2 --with pytest-xdist==3.8.0 --with pytest-asyncio==1.3.0 python -m pytest -q tests/agent/test_auxiliary_client.py -k "TestBuildCallKwargsZaiVision"
2. Run touched-file lint:
   • uv run --with ruff==0.15.10 ruff check agent/auxiliary_client.py tests/agent/test_auxiliary_client.py
3. Reproduce with a GLM vision custom endpoint (open.bigmodel.cn/api.z.ai) and verify requests no longer include max_tokens and no longer fail with 1210 due to this parameter.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS 15.x

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

• Targeted test output: 3 passed in 3.78s
• git diff --check: clean