fix(tool-config): respect question permission from OPENCODE_CONFIG_CONTENT

applyToolConfig() unconditionally set question permission based only on
OPENCODE_CLI_RUN_MODE, ignoring the question:deny already configured via
OPENCODE_CONFIG_CONTENT. This caused agents to hang in headless environments
(e.g. Maestro Auto Run) where the host sets question:deny but does not
know about the plugin-internal OPENCODE_CLI_RUN_MODE variable.

Read permission.question from OPENCODE_CONFIG_CONTENT and give it highest
priority: config deny > CLI run mode deny > default allow.

src/plugin-handlers/tool-config-handler.ts

@@ -3,6 +3,17 @@ import { getAgentDisplayName } from "../shared/agent-display-names";
 
 type AgentWithPermission = { permission?: Record<string, unknown> };
 
+function getConfigQuestionPermission(): string | null {
+  const configContent = process.env.OPENCODE_CONFIG_CONTENT;
+  if (!configContent) return null;
+  try {
+    const parsed = JSON.parse(configContent);
+    return parsed?.permission?.question ?? null;
+  } catch {
+    return null;
+  }
+}
+
 function agentByKey(agentResult: Record<string, unknown>, key: string): AgentWithPermission | undefined {
   return (agentResult[key] ?? agentResult[getAgentDisplayName(key)]) as
     | AgentWithPermission
@@ -32,7 +43,11 @@ export function applyToolConfig(params: {
   };
 
   const isCliRunMode = process.env.OPENCODE_CLI_RUN_MODE === "true";
-  const questionPermission = isCliRunMode ? "deny" : "allow";
+  const configQuestionPermission = getConfigQuestionPermission();
+  const questionPermission =
+    configQuestionPermission === "deny" ? "deny" :
+    isCliRunMode ? "deny" :
+    "allow";
 
   const librarian = agentByKey(params.agentResult, "librarian");
   if (librarian) {