Topic_listener: allow multi-topic messages

[jlecoeur]

This PR allows uORB messages with multi topics (defined with the syntax # TOPICS ...) in topic_listener.

Now it is possible to do:

pxh> listener actuator_controls_0

Here is the list of topics that where not working with topic_listener:

• actuator_controls_0
• actuator_controls_1
• actuator_controls_2
• actuator_controls_3
• actuator_controls_virtual_fw
• actuator_controls_virtual_mc
• offboard_mission
• onboard_mission
• vehicle_attitude_groundtruth
• vehicle_vision_attitude
• mc_virtual_attitude_setpoint
• fw_virtual_attitude_setpoint
• vehicle_global_position_groundtruth
• vehicle_local_position_groundtruth
• vehicle_vision_position
• mc_virtual_rates_setpoint
• fw_virtual_rates_setpoint

I also fixed the generated cpp code because the first message in the list (currently commander_state) was not printed by topic_listener.

closes #6607
@mhkabir

[jlecoeur]

There are too many topics now ! 🥇

/home/travis/build/PX4/Firmware/build/posix_sitl_default-clang/src/systemcmds/topic_listener
/topic_listener.cpp:164:5: error: function 'listener_main' exceeds recommended 
size/complexity thresholds [google-readability-function-size,-warnings-as-errors]

note: 4143 statements (threshold 4000)
note: 616 branches (threshold 600)

@dagar Can we raise that complexity limit, or white-list the autogenerated files?

[TSC21]

@jlecoeur try to increase it and check for the stack and ram usage to see if it sustainable.

[dagar]

No problem, the complexity was supposed to be a cap, but doesn't really make sense with generated code like this. You can increase it here https://github.com/PX4/Firmware/blob/master/.clang-tidy

[jlecoeur]

In the end I modified the script to generate one short function per topic.
4000 is already way larger than the recommended value

[jlecoeur]

Regarding memory usage @TSC21 that should be fine: the structures are allocated on the stack when needed, and they are not larger than before. Also listener subscribes to a single topic, then unsubscribes. I simply expect a bit larger firmware because more code gets generated.

[bkueng]

Thanks!
This is the firmware size for v4:

before:
1456420    4133   24972 1485525  16aad5 build/px4fmu-v4_default/nuttx_px4fmu-v4_default.elf
before w/o listener:
1412532    4133   24972 1441637  15ff65 build/px4fmu-v4_default/nuttx_px4fmu-v4_default.elf
this PR:
1466100    4133   24972 1495205  16d0a5 build/px4fmu-v4_default/nuttx_px4fmu-v4_default.elf

So FW size is increased by 9680 bytes, but the listener was already big (43888 bytes).