threadshare: block_on panics because it is running in a Context

I have been trying to fix this on my end but my brain isn't working anymore :(

Consider the following pipeline:

gst-launch-1.0 audiotestsrc ! alawenc ! rtppcmapay ! ts-jitterbuffer ! rtppcmadepay ! alawdec ! fakesink

This is a simplified scenario, my actual use case has a sender and receiver pipeline, but it can be reproduced with this single pipeline too.

stack.txt

Here, audiodecoder (from its streaming thread) constructs and sends a latency query upstream (frame 30). The jitterbuffer receives it and attempts to block_on in its src_query handler, which panics as it is indeed running in its own context (see the pop_and_push at frame 55).

I don't know how to fix this :)