Fix node database ATC rail bug

After successfully registering newly-placed ATC rails in the nodedb, the rail's after_place_node() callback
(local apn_func()) immediately cleared it again by passing a "node" object that was actually a player object
and thus had no "name" or "param2" to ndb.update(), which it handled how it should not in this situation.
Since those ndb.update calls were unnecessary anyways, I removed them completely.

advtrains/atc.lua

@@ -78,8 +78,8 @@ end
 
 --nodes
 local idxtrans={static=1, mesecon=2, digiline=3}
-local apn_func=function(pos, node)
-	advtrains.ndb.update(pos, node)
+local apn_func=function(pos)
+	-- FIX for long-persisting ndb bug: there's no node in parameter 2 of this function!
 	local meta=minetest.get_meta(pos)
 	if meta then
 		meta:set_string("infotext", attrans("ATC controller, unconfigured."))

advtrains/nodedb.lua

@@ -153,10 +153,13 @@ function ndb.update(pos, pnode)
 			nid=#ndb_nodeids+1
 			ndb_nodeids[nid]=node.name
 		end
-		ndbset(pos.x, pos.y, pos.z, (nid * 4) + (l2b(node.param2 or 0)) )
-		--atprint("nodedb: updating node", pos, "stored nid",nid,"assigned",ndb_nodeids[nid],"resulting cid",ndb_nodes[hash])
+		local resid = (nid * 4) + (l2b(node.param2 or 0))
+		ndbset(pos.x, pos.y, pos.z, resid )
+		--atdebug("nodedb: updating node", pos, "stored nid",nid,"assigned",ndb_nodeids[nid],"resulting cid",resid)
+		advtrains.invalidate_all_paths(pos)
 	else
 		--at this position there is no longer a node that needs to be tracked.
+		--atdebug("nodedb: updating node", pos, "cleared")
 		ndbset(pos.x, pos.y, pos.z, nil)
 	end
 end

advtrains_luaautomation/active_common.lua

@@ -12,7 +12,6 @@ function ac.save()
 end
 
 function ac.after_place_node(pos, player)
-	advtrains.ndb.update(pos)
 	local meta=minetest.get_meta(pos)
 	meta:set_string("formspec", ac.getform(pos, meta))
 	meta:set_string("infotext", "LuaAutomation component, unconfigured.")