[fmrico]

Needs EasyNavigation/EasyFleet#5 (capability API, TF relay, fleet obstacles, easyfleet_bt_nodes). Two commits.

1. Playgrounds and Zenoh, per-robot TF with a fleet RViz, READMEs and fixes

• Simulations from easynav_playgrounds (simulation_launch.yaml in each package), and rmw_zenoh_cpp by default with its router.
• Each robot keeps its TF under its namespace; navigation_manager relays only odom, base_footprint and base_link to /tf, for a central RViz with each robot's base and path.
• Launch arguments to choose each robot's EasyNav, navigation and navigation manager parameters, its own RViz and the mission.
• Python fake capabilities for simple_api_py; the fastdds (DDS Router) and farm (Summit, GPS, NavMap) deployments, listed in the README but never committed.
• Navigation capability: its executor is only spun from on_execute(), and a cancel waits for EasyNav's answer, so it no longer disturbs the next goal.
• The collaboration mission said every phase succeeded even when a robot's goal ended CANCELED; it now says which robot missed which waypoint.
• A README per deployment.

2. Warehouse A deployments and a PlanSys2 fleet

• easyfleet_easynav_deployment (one robot) and easyfleet_easynav_collaboration_deployment (two) in Warehouse A of large_warehouse_world:
  • a 0.2 m costmap of what the lidar sees at its height (rack legs, packages, walls) and routes along every rack aisle, the cross corridors and to the docks, generated from the world by scripts/sdf_to_costmap.py and scripts/aisle_routes.py;
  • AMCL with reseeding (fixed in Costmap AMCL: guard the particles and bound the reseed noise easynav_plugins#145), a 30 m lidar (at 12 m, 0–39 of 360 rays hit anything in the open areas; at 30 m, 200+) and lidar_target on each Kobuki (Kobuki: optional lidar target easynav_playgrounds#28).
• easyfleet_plansys2_warehouse_deployment: four Kobukis, two scouts (perception) and two carriers (manipulation), driven by PlanSys2 with Kairos. The controller builds the PDDL problem from the capabilities the robots announce and reaches the mission's goals in turn; each action's behavior tree calls the robot's capabilities with easyfleet_bt_nodes.

Tested: every package's tests and linters pass. In simulation, one Kobuki reaches a rack aisle 0.15 m from where it believes it is, and the four Kobukis deliver the first goal's two packages in 657 s, all of them within 0.35 m of their estimate along 12 minutes.

The warehouse world comes from juanscelyg/large_warehouse_world (its changes merged in juanscelyg/large_warehouse_world#1).