feat(providers): add Bailian GLM 5.2 fast preview

[guima-why]

Summary

• add glm-5.2-fast-preview to the standard Alibaba Cloud Bailian provider catalog only
• route the exact model ID to DashScope and register its thinking capabilities
• configure the documented 1,048,576-token context window and 131,072-token output limit
• add fallback and telemetry coverage
• extend provider, routing, thinking, wire-format, context, and telemetry tests

Why

Alibaba Cloud Bailian now exposes glm-5.2-fast-preview as a faster preview variant of GLM 5.2. The model uses Bailian's GLM 5.2 hybrid-thinking protocol and is not listed for the Token Plan endpoint.

Impact

Users can select the preview model from the standard Bailian provider while the existing default model remains unchanged. The model is treated as text-only and falls back to glm-5.2 on eligible degradation paths.

Validation

• make lint
• focused provider/context/telemetry tests: 339 passed
• full parallel suite: 13,276 passed, 2 unrelated process-timing tests passed on isolated rerun
• git diff --check