Refactor power_main.cpp and improve safety checks to thursters

[asiancooki]

sending 1500us to thursters at shutdown from thurster ks or any shutdown

Refactored error_loop and callback functions, added initialization for microseconds array on reset, and ensured safety before destroying entities.

In power_setup() (line ~242): Initializes the microseconds array immediately after the message buffer initialization. This ensures that when the Teensy resets (kill switch pulled/reinserted), the thrusters start at neutral (1500) before any ROS commands arrive.​

In create_entities() (line ~155): Resets the microseconds array when ROS entities are created/reconnected. This handles the case where the ROS node on your host computer didn't die but the Teensy did—old commands from the still-running node won't override the safe 1500 value until new commands arrive.​

In destroy_entities() (line ~171): Sets thrusters to 1500 before calling offCommand for extra safety during disconnection.

[asiancooki]

Tweaked the Teensy power node to be safer and easier to debug.

Added a propulsion “watchdog” that forces all thrusters to 1500 µs if no /propulsion/microseconds command is received for a while (currently 2 s, configurable in code).

Watchdog is non-blocking and only kicks in once per timeout window, so it does not fight normal commands or cut short 1 s maneuvers.

When a new propulsion message comes in, the watchdog is reset and normal control resumes.

Added a /power/state string topic to echo the micro-ROS connection state (WAITING_AGENT, AGENT_CONNECTED, etc.) for easier debugging on the host.

Kept all existing publishers, timers, and thruster logic intact; changes are mostly wiring in the timeout + state echo around the existing state machine.

Tested with dry runs by:

Sending step commands on /propulsion/microseconds and checking the thrusters follow for the full duration.

Stopping commands and confirming the system falls back to 1500 µs on all channels.