[Bug] the leak test flakiness fixed by delay in test. Proper fix needed

[milyin]

Describe the bug

#906 fixes test failure on CI (irreproducable locally)

19/20 Test #19: leak_z_leak_pub_sub_test ......................***Failed    8.80 sec
Checking /home/runner/work/zenoh-c/zenoh-c/target/release/tests/z_leak_pub_sub_test for memory leaks
thread 'rx-0' panicked at /home/runner/.cargo/git/checkouts/zenoh-cc237f2570fab813/27bb8df/commons/zenoh-runtime/src/lib.rs:212:32:
The hashmap should contains app after initialization
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
/home/runner/work/zenoh-c/zenoh-c/tests/run_leak_check.sh: line 5: 16604 Aborted                 (core dumped) valgrind --leak-check=full --num-callers=50 --log-file="$1.leaks.log" $1

This is not very important functionality for end user: the function which supposedly fails zc_stop_z_runtime is necessary only for verification that no memory leaks occurs in zenoh. But it's better to have a proper fix someday

To reproduce

System info

[DenisBiryukov91]

it is not zc_stop_z_runtime that fails, it is more, like there are still tasks that try to access ZenohRuntime that did not finish prior to zc_stop_z_runtime call.