Support per-tool MCP output limits (#41421)

## What changed

- Add a positive `output_token_limit` setting to each entry under an MCP server's `tools` configuration.
- Apply the most restrictive limit when plugin and user policies overlap, while keeping approval policy independent.
- Carry the effective MCP output budget in conversation history so tool output, post-tool hook responses, and resumed sessions use the same truncation limit.

## Testing

- Cover configuration parsing, serialization, schema validation, and plugin policy merging.
- Cover MCP output below and above the configured limit, post-tool hook responses, and session resume.

GitOrigin-RevId: d0beb4fca9ba6055d9e1d31c137373b465d50d61

codex-rs/core-plugins/src/loader.rs

@@ -974,6 +974,7 @@ fn apply_plugin_mcp_server_policy(config: &mut McpServerConfig, policy: &PluginM
         if let Some(approval_mode) = tool_policy.approval_mode {
             tool_config.approval_mode = Some(approval_mode);
         }
+        tool_config.restrict_output_token_limit(tool_policy.output_token_limit);
     }
 }
 
@@ -1449,7 +1450,7 @@ pub fn apply_configured_plugin_mcp_server_policies(
                 }
             }
             for (tool_name, tool_policy) in &policy.tools {
-                if tool_policy.approval_mode.is_some() {
+                if tool_policy.approval_mode.is_some() || tool_policy.output_token_limit.is_some() {
                     server.tools.entry(tool_name.clone()).or_default();
                 }
             }
@@ -1467,6 +1468,12 @@ pub fn apply_configured_plugin_mcp_server_policies(
                             .restrict_to(approval_mode),
                     );
                 }
+                tool_config.restrict_output_token_limit(
+                    policy
+                        .tools
+                        .get(tool_name)
+                        .and_then(|tool_policy| tool_policy.output_token_limit),
+                );
             }
         }
     }