qemu_cortex_a53_smp and qemu_x86_64 failed in tests/kernel/condvar/condvar while enabling for SMP

[enjiamai]

Describe the bug
Our new added test case try to remove the limit of CONFIG_MP_NUM_CPUS=1 for tests/kernel/condvar/ testing. Please refer to PR #33547.
There are two test case failed while removing the CONFIG_MP_NUM_CPUS=1
(1) test_condvar_usecase_broadcast
(2) test_condvar_usecase_signal

To Reproduce
Steps to reproduce the behavior:

1. Edit the "tests/kernel/condvar/condvar_api/prj.conf", then comment or remove the line CONFIG_MP_NUM_CPUS=1.

2. Run command: west build tests/kernel/smp -p auto -b qemu_cortex_a53_smp(or qemu_x86_64) -t run
   or twister -T tests/kernel/smp -p qemu_cortex_a53(or -p qemu_x86_64(or qemu_x86_64)

Test cases failed on qemu_cortex_a53_smp and qemu_x86_64

3. It will hang up the test case mentioned above. See the error log below, please:

Expected behavior
Test cases are pass as expected.

===================================================================
START - test_condvar_usecase_signal
Starting watch_count: thread 1
 PASS - test_condvar_usecase_signal in 2.102 seconds
===================================================================
START - test_condvar_usecase_broadcast
Starting watch_count: thread 1
 PASS - test_condvar_usecase_broadcast in 2.100 seconds
===================================================================

Impact
Not clear.

Logs and console output
Error log looks like this:

===================================================================
START - test_condvar_usecase_signal
Starting watch_count: thread 1

or

===================================================================
START - test_condvar_usecase_broadcast
Starting watch_count: thread 1

Environment (please complete the following information):

• OS: Linux
• Toolchain Zephyr SDK-0.12.3
• Commit SHA 911ba33

Additional context

[enjiamai]

Hi @nashif , This is not a bug of condvar's code. It has something to do with test cases, could you please take a look at #33547? I put the fix in that PR, thanks!