test_ddp.py hangs at test_ddp_configure_ddp

Nobody has claimed this yet.

### Bug description

When I try to run single test on multi-gpu device, it hangs at specific test case `test_ddp_configure_ddp` https://github.com/Lightning-AI/pytorch-lightning/blob/6f6c07dddfd68717f0b765a63d05a937b8508e15/tests/tests_pytorch/strategies/test_ddp.py#L142.

I've found that this case is related to `torch.distributed.init_process_group` . https://github.com/Lightning-AI/pytorch-lightning/blob/6f6c07dddfd68717f0b765a63d05a937b8508e15/src/lightning/fabric/utilities/distributed.py#L258

I have 4 gpus on my device, so the `world_size` send to `init_process_group` is 4, but the way I ran the test seems not include multiprocessing, so `init_process_group` just hangs for device to be ready.

When I set `CUDA_VISIBLE_DEVICES` to single device, it works fine. I'm not sure is this a bug or I've used the wrong command to run test. If so. how should I run this test?

### What version are you seeing the problem on?

v2.1

### How to reproduce the bug

```python
On multi-gpu environment:

pytest -v tests/tests_pytorch/strategies/test_ddp.py -k test_ddp_configure_ddp
```

### Error messages and logs

```
# Error messages and logs here please
no Error message, just hang up for a long time(default args: 1800 seconds)
```

### Environment

Current environment

* CUDA:
- GPU:
- Tesla V100-SXM2-16GB
- Tesla V100-SXM2-16GB
- Tesla V100-SXM2-16GB
- Tesla V100-SXM2-16GB
- available: True
- version: 12.2
* Lightning:
- lightning: 2.1.0
- lightning-utilities: 0.10.1
- pytorch-lightning: 2.2.1
- torch: 2.3.0a0+gitbfa71b5
- torchmetrics: 1.3.1
- torchvision: 0.18.0a0+423a1b0
* Packages:
- aiohttp: 3.9.4rc0
- aiosignal: 1.3.1
- annotated-types: 0.6.0
- astunparse: 1.6.3
- async-timeout: 4.0.3
- attrs: 23.2.0
- audioread: 3.0.1
- certifi: 2023.11.17
- cffi: 1.16.0
- charset-normalizer: 3.3.2
- contourpy: 1.2.0
- cycler: 0.12.1
- decorator: 5.1.1
- dllogger: 1.0.0
- exceptiongroup: 1.2.0
- expecttest: 0.2.1
- filelock: 3.13.1
- fonttools: 4.49.0
- frozenlist: 1.4.1
- fsspec: 2023.12.2
- hypothesis: 6.97.3
- idna: 3.6
- inflect: 7.0.0
- iniconfig: 2.0.0
- jinja2: 3.1.3
- joblib: 1.3.2
- kiwisolver: 1.4.5
- librosa: 0.8.1
- lightning: 2.1.0
- lightning-utilities: 0.10.1
- llvmlite: 0.42.0
- markupsafe: 2.1.4
- matplotlib: 3.8.3
- mpmath: 1.3.0
- multidict: 6.0.5
- networkx: 3.2.1
- numba: 0.59.0
- numpy: 1.23.1
- optree: 0.10.0
- packaging: 23.2
- pillow: 10.2.0
- pip: 22.2.2
- platformdirs: 4.2.0
- pluggy: 1.4.0
- pooch: 1.8.1
- psutil: 5.9.8
- pycparser: 2.21
- pydantic: 2.6.3
- pydantic-core: 2.16.3
- pyparsing: 3.1.2
- pytest: 8.1.1
- python-dateutil: 2.9.0.post0
- pytorch-lightning: 2.2.1
- pyyaml: 6.0.1
- requests: 2.31.0
- resampy: 0.4.3
- scikit-learn: 1.4.1.post1
- scipy: 1.12.0
- setuptools: 63.2.0
- six: 1.16.0
- sortedcontainers: 2.4.0
- soundfile: 0.12.1
- sympy: 1.12
- tabulate: 0.9.0
- threadpoolctl: 3.3.0
- tomli: 2.0.1
- torch: 2.3.0a0+gitbfa71b5
- torchmetrics: 1.3.1
- torchvision: 0.18.0a0+423a1b0
- tqdm: 4.66.2
- types-dataclasses: 0.6.6
- typing-extensions: 4.9.0
- unidecode: 1.3.8
- urllib3: 2.2.0
- wheel: 0.42.0
- yarl: 1.9.4
* System:
- OS: Linux
- architecture:
- 64bit
- ELF
- processor: x86_64
- python: 3.10.8
- release: 5.11.0-27-generic
- version: #29~20.04.1-Ubuntu SMP Wed Aug 11 15:58:17 UTC 2021

### More info

_No response_

cc @lantiga @borda

Start with tests/tests_pytorch/strategies/test_ddp.py at test_ddp_configure_ddp and the distributed initialization path in src/lightning/fabric/utilities/distributed.py. Run the reported pytest command on a multi-GPU environment and compare it with a single visible device. Done means the test no longer hangs when invoked as documented, or the correct multiprocessing command and test guidance are documented.

Written by the indexing model from the issue text.