ocm: Add Responses WebSocket API proxy and fix client config docs

Support the OpenAI Responses WebSocket API (`wss://.../v1/responses`)
for bidirectional frame proxying with usage tracking.
Fix Codex CLI client config examples to use profiles and correct flags.

Update openai-go v3.24.0 → v3.26.0.

docs/configuration/service/ocm.zh.md

@@ -114,14 +114,18 @@ TLS 配置，参阅 [TLS](/zh/configuration/shared/tls/#inbound)。
 [model_providers.ocm]
 name = "OCM Proxy"
 base_url = "http://127.0.0.1:8080/v1"
-wire_api = "responses"
-requires_openai_auth = false
+supports_websockets = true
+
+[profiles.ocm]
+model_provider = "ocm"
+# model = "gpt-5.4"              # 如果最新模型尚未公开发布
+# model_reasoning_effort = "xhigh"
 ```
 
 然后运行：
 
 ```bash
-codex --model-provider ocm
+codex --profile ocm
 ```
 
 ### 带身份验证的示例
@@ -159,13 +163,17 @@ codex --model-provider ocm
 [model_providers.ocm]
 name = "OCM Proxy"
 base_url = "http://127.0.0.1:8080/v1"
-wire_api = "responses"
-requires_openai_auth = false
+supports_websockets = true
 experimental_bearer_token = "sk-alice-secret-token"
+
+[profiles.ocm]
+model_provider = "ocm"
+# model = "gpt-5.4"              # 如果最新模型尚未公开发布
+# model_reasoning_effort = "xhigh"
 ```
 
 然后运行：
 
 ```bash
-codex --model-provider ocm
+codex --profile ocm
 ```