recv: optional wait_seconds parameter, capped at 60s

AgentRequest::Recv and ManagerRequest::Recv grow an optional
wait_seconds field (default None → 30s, capped at 60s server-side).
agent_server / manager_server clamp via recv_timeout(). MCP tool
schemas advertise the param so claude can pick its own poll window
— useful when an agent wants to throttle wakes without entering a
distinct nap state.

both harness loops still pass None, keeping the existing 30s
default behaviour for system-level Recvs.

hive-sh4re/src/lib.rs

@@ -166,8 +166,13 @@ pub struct InboxRow {
 pub enum AgentRequest {
     /// Send a message to another agent.
     Send { to: String, body: String },
-    /// Pop one pending message from this agent's inbox.
-    Recv,
+    /// Pop one pending message from this agent's inbox. Long-polls
+    /// up to `wait_seconds` (capped at 60s server-side, default 30s
+    /// when None) before returning `Empty`.
+    Recv {
+        #[serde(default)]
+        wait_seconds: Option<u64>,
+    },
     /// Non-mutating: how many pending messages are addressed to me?
     /// Used by the harness to render a status line after each tool call.
     Status,
@@ -274,7 +279,12 @@ pub enum ManagerRequest {
         to: String,
         body: String,
     },
-    Recv,
+    /// Same shape as `AgentRequest::Recv` — caller-tunable long-poll
+    /// duration, capped at 60s server-side, default 30s when None.
+    Recv {
+        #[serde(default)]
+        wait_seconds: Option<u64>,
+    },
     /// Non-mutating: pending message count, used to render a status line
     /// after each MCP tool call (mirrors `AgentRequest::Status`).
     Status,