Cannot reach the endpoint configured on macOS

**Describe the bug**
I am on macOS BigSur 11.3.1, and I have to use some tricky ways to install python 3.6 and 3.7 to run `funcx-endpoint` command without warnings. Otherwise, there will warnings as the following

> /Users/apple/.pyenv/versions/3.9.0/lib/python3.9/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 6 leaked semaphore objects to clean up at shutdown
warnings.warn('resource_tracker: There appear to be %d '

After I installed python 3.6 on Big Sur following the tutorial https://dev.to/kojikanao/install-python-3-7-3-6-and-3-5-on-bigsure-with-pyenv-3ij2, the warning is gone (probably because there is no resource_tracker in python 3.6). When I start an endpoint, it cannot be reached (I mean the endpoint id is not recognized). I did exactly the same steps on a virtual Linux system, there is no such problem.

**To Reproduce**
Steps to reproduce the behavior, for e.g:
1. Make sure you are on macOS Big Sur 11.3.1 (maybe 11.2 also has the problem)
2. Use `pyenv` to install python 3.9 normally
3. `pip install funcx-endpoint` in the pyenv virtualenv
4. `funcx-endpoint configure default` to configure an endpoint
5. `funcx-endpoint start default` and you should see the warning, and the endpoint is not started correctly
6. `funcx-endpoint stop default` and you will see some exception.
7. Try installing python 3.7 to avoid the above problem.
8. The exception no longer exists, but the endpoint cannot be reached.

**Environment**
- OS: macOS Bigsur 11.3.1

**The Screenshot**

Start with the funcx-endpoint configure, start, and stop commands and reproduce the behavior on macOS Big Sur with Python 3.9 and 3.7. Compare the startup warning, stop exception, and endpoint reachability with the Linux result; done means the endpoint starts, is recognized, and stops without the reported failures.

Written by the indexing model from the issue text.