Fix oled by using thread safe run_coroutine_threadsafe

[ArendJan]

the ros service is started on a different thread than the asyncio loop.
When using the normal loop.run_until_complete() function, both threads join in and there might be race conditions and the oled function might break faster.
With run_corotine_threadsafe, the function is started on the asyncio thread, without the ros service thread joining in and 'helping'.

Tested it with this code:

from mirte_robot import robot
mirte=robot.createRobot()
for count2 in range(100000000):
  mirte.setOLEDText('left', ('Hi' + str(count2)))

Worked for at least 10 minutes at 5Hz.

[ArendJan]

When it crashes, it does not crash the whole telemetrix node, but only the communication with the oled.
This is either an i2c issue or oled issue. Restarting the communication with the oled or restarting the i2c might fix this.

I think that this is not an thing to solve for v0.1, but something to look into for v0.2

[mklomp]

Nice that it keeps running for this long. Agreed that we might need to have a more thorough look at it for v0.2.

We also might have a look at aiorospy. Initially this was used everywhere. Not sure why I removed the, except for the OLED. This might also be something to look into. Prefereably I do not use aiorospy (extra dependence, and not (yet) ported to ROS2). But as mentioned, I have no clue why I removed it almost everywhere else.