feat(providers): GLM-5.2 native reasoning_effort controls (#58884)

Port from Kilo-Org/kilocode#11555: GLM-5.2 exposes a native
reasoning_effort knob with two enabled levels (high / max) on its
OpenAI-compatible endpoints. Previously the zai profile (direct Z.AI
/api/paas/v4) used the base ProviderProfile and emitted nothing, and the
OpenCode Go profile only handled Kimi K2 / DeepSeek — so a user's effort
preference for GLM-5.2 was silently dropped on both routes.

- zai: ZaiProfile maps effort onto high/max (xhigh/max -> max, lower -> high)
- opencode-go: same mapping for GLM-5.2, alongside existing Kimi/DeepSeek
- alias spellings recognized (glm-5.2 / glm-5-2 / glm-5p2, vendor-prefixed)
- disabled / no effort leaves the server default untouched

plugins/model-providers/opencode-zen/__init__.py

@@ -31,6 +31,12 @@ def _is_deepseek_thinking_model(model: str | None) -> bool:
     return m == "deepseek-reasoner"
 
 
+def _is_glm_5_2_model(model: str | None) -> bool:
+    """Detect GLM-5.2 across alias spellings (glm-5.2 / glm-5-2 / glm-5p2)."""
+    m = _flat_model_name(model)
+    return any(token in m for token in ("glm-5.2", "glm-5-2", "glm-5p2"))
+
+
 class OpenCodeGoProfile(ProviderProfile):
     """OpenCode Go - model-specific reasoning controls."""
 
@@ -55,6 +61,21 @@ class OpenCodeGoProfile(ProviderProfile):
         extra_body: dict[str, Any] = {}
         top_level: dict[str, Any] = {}
 
+        if _is_glm_5_2_model(model):
+            # GLM-5.2 on OpenCode Go uses its native OpenAI-compatible
+            # reasoning_effort knob, which has exactly two enabled levels:
+            # high and max. Map Hermes' richer scale onto those; leave the
+            # server default alone when reasoning is disabled or unset.
+            if not isinstance(reasoning_config, dict):
+                return extra_body, top_level
+            if reasoning_config.get("enabled") is False:
+                return extra_body, top_level
+            effort = (reasoning_config.get("effort") or "").strip().lower()
+            if not effort or effort == "none":
+                return extra_body, top_level
+            top_level["reasoning_effort"] = "max" if effort in {"xhigh", "max"} else "high"
+            return extra_body, top_level
+
         if _is_kimi_k2_model(model):
             # Kimi K2 on OpenCode Go uses Moonshot's native wire shape:
             # extra_body.thinking (binary toggle) + top-level reasoning_effort