Cache architecture: identity fix, cache merge, parent-change flush

Identity cache: remove TTL-based expiry (60s TTL broke active sessions
after expiry since handle_tun_outbound checks identity_cache before
session table). Replace with LRU-only eviction bounded by configurable
identity_size (default 10K). Lookup now touches timestamp for LRU
freshness.

Cache merge: unify coord_cache and route_cache into single coordinate
cache. Both stored NodeAddr→TreeCoordinate; the layer distinction was
conceptual, not functional. Discovery-sourced entries now get the same
TTL+refresh treatment as session-sourced entries. Simplifies
find_next_hop() to single cache lookup.

Parent-change flush: clear coord_cache after recompute_coords() in both
parent-switch paths of handle_tree_announce(). Stale coordinates after
tree reconvergence cause dead-end routing that's more expensive than
re-discovery.

Tested: 493 unit tests passed, clippy clean, Docker mesh 20/20,
Docker chain 6/6.

src/node/handlers/discovery.rs

@@ -92,7 +92,7 @@ impl Node {
     /// Processing steps:
     /// 1. Decode and validate
     /// 2. Check recent_requests to determine if we originated or are forwarding
-    /// 3. If originator: cache target_coords in route_cache
+    /// 3. If originator: cache target_coords in coord_cache
     /// 4. If transit: reverse-path forward to from_peer
     pub(in crate::node) async fn handle_lookup_response(
         &mut self,
@@ -139,7 +139,7 @@ impl Node {
             );
 
             let target = response.target;
-            self.route_cache.insert(
+            self.coord_cache.insert(
                 target,
                 response.target_coords,
                 now_ms,
@@ -149,7 +149,7 @@ impl Node {
             self.pending_lookups.remove(&target);
 
             // If we have pending TUN packets for this target, retry session
-            // initiation. The route_cache now has coords, so find_next_hop()
+            // initiation. The coord_cache now has coords, so find_next_hop()
             // should succeed.
             if self.pending_tun_packets.contains_key(&target) {
                 self.retry_session_after_discovery(target).await;
@@ -261,7 +261,7 @@ impl Node {
     /// Creates a LookupRequest and floods it to all peers. The originator
     /// does NOT record the request_id in recent_requests, so when the
     /// response arrives, it's recognized as "our request" and the
-    /// target's coordinates are cached in route_cache.
+    /// target's coordinates are cached in coord_cache.
     pub(in crate::node) async fn initiate_lookup(&mut self, target: &NodeAddr, ttl: u8) {
         let origin = *self.node_addr();
         let origin_coords = self.tree_state().my_coords().clone();

src/node/handlers/session.rs

@@ -497,9 +497,6 @@ impl Node {
         if let Some(coords) = self.coord_cache.get(dest, now_ms) {
             return coords.clone();
         }
-        if let Some(cached) = self.route_cache.get(dest) {
-            return cached.coords().clone();
-        }
         // Fallback: use our own coordinates. The SessionSetup dest_coords
         // field cannot be empty (wire format requires ≥1 entry). Using our
         // own coords is safe — transit routers will still cache them, and
@@ -675,7 +672,7 @@ impl Node {
     /// Retry session initiation after discovery provided coordinates.
     ///
     /// Called when a LookupResponse arrives and we have pending TUN packets
-    /// for the discovered target. The route_cache now has coords, so
+    /// for the discovered target. The coord_cache now has coords, so
     /// `find_next_hop()` should succeed and the SessionSetup can be sent.
     pub(in crate::node) async fn retry_session_after_discovery(&mut self, dest_addr: NodeAddr) {
         // Look up the destination's public key from the identity cache