fix: fix p2p connection to use loop call soon threadsafe

[mark-neuracore]

Bugfixes

• p2p gives "no current event loop in this thread" error

⚠️  Failed to call log_parallel_gripper_target_open_amounts to Neuracore: There is no current event loop in thread 'Thread-2 (joint_state_thread)'.
/home/mark/neuracore_ws/example_elephant_robotics/examples/2_collect_teleop_data_with_neuracore.py:102: RuntimeWarning: coroutine 'RTCSctpTransport._data_channel_flush' was never awaited
print(f"\n⚠️  Failed to call {name} to Neuracore: {e}")
RuntimeWarning: Enable tracemalloc to get the object allocation traceback

• This was due to adding event sources needed to use an event loop, which if called from a different thread (that never initialised an event loop) causes this error. This method uses self.loop instead.

[github-actions]

✅ PR source branch is valid

• Source: fix/fix-p2p-connection-to-use-loop-call-soon-threadsafe
• Target: develop

[CougarTasker]

Issue: This solution feels like we are just smothing over the issue. Mixing blocking and non-blocking code like this raises alarm-bells for me.

Suggestion: I suspect there is a more elegant approach to provide the event loop context to the required code.

Question: Does this issue apprear for other senesor types that are not parallel grippers, If not why?

[mark-neuracore]

@CougarTasker found this way to avoid modifying aiortc.

still happy to raise a pr with our previous solution to pass the self.loop in aiortc

[CougarTasker]

LGTM, nice tests!

[CougarTasker]

nit: These callback functions can look a little neater if inline