Cannot make test case unused parameters for proposed strategy in PR

Nobody has claimed this yet.

### Bug description

Hi, I'm currently trying to pass the PR #20936 and have a problem while writing a certain test case.

Lately, I added some test cases in that PR and test them successfully except for below test case:

```
L141 - L165 of tests/tests_pytorch/strategies/test_multi_model_ddp.py in above PR

# class GeneratorWithUnused(Generator):
# def __init__(self, latent_dim, img_shape):
# super().__init__(latent_dim, img_shape)
# self.unused = torch.nn.Linear(latent_dim, latent_dim)

# def forward(self, z):
# z = self.unused(z)
# z = z.detach()
# return super().forward(z)

# class UnusedParametersModel(GenerationModel):
# def __init__(self):
# super().__init__()
# self.generator = GeneratorWithUnused(latent_dim=128, img_shape=(1, 28, 28))

# def training_step(self, batch, batch_idx):
# return super().training_step(batch, batch_idx)

# @RunIf(standalone=True)
# def test_find_unused_parameters_multi_model_ddp_raises():
# trainer = Trainer(accelerator="cpu", devices=1, strategy=MultiModelDDPStrategy(), max_steps=2, logger=False)
# with pytest.raises(RuntimeError, match="It looks like your LightningModule has parameters that were not used in"):
# trainer.fit(UnusedParametersModel())
```

I write that code based on the `tests/tests_pytorch/strategies/test_ddp_integration.py` but I fail to satisfy the original one's intention; test whether there is no unused parameter during the training in the model.

Is there anyone who can help me to write this kind of test case?

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

master

### Reproduced in studio

_No response_

### How to reproduce the bug

```python

```

### Error messages and logs

```
# Error messages and logs here please
```

### Environment

Current environment

```
aiohappyeyeballs 2.6.1
aiohttp 3.11.14
aiosignal 1.3.2
annotated-types 0.7.0
antlr4-python3-runtime 4.9.3
anykeystore 0.2
apex 0.9.10.dev0
attrs 25.3.0
certifi 2025.1.31
charset-normalizer 3.4.1
click 8.1.8
cryptacular 1.6.2
decorator 4.4.2
defusedxml 0.7.1
docker-pycreds 0.4.0
facenet-pytorch 2.6.0
filelock 3.18.0
flow-vis 0.1
frozenlist 1.5.0
fsspec 2025.3.0
gitdb 4.0.12
GitPython 3.1.44
greenlet 3.0.3
h5py 3.11.0
hupper 1.12.1
idna 3.10
Jinja2 3.1.6
lightning-utilities 0.14.1
MarkupSafe 3.0.2
mpmath 1.3.0
multidict 6.2.0
munkres 1.1.4
munkres 1.1.4
natsort 8.4.0
natsort 8.4.0
networkx 3.4.2
numpy 1.26.4
nvidia-cublas-cu12 12.1.3.1
nvidia-cuda-cupti-cu12 12.1.105
nvidia-cuda-nvrtc-cu12 12.1.105
nvidia-cuda-runtime-cu12 12.1.105
nvidia-cudnn-cu12 8.9.2.26
nvidia-cufft-cu12 11.0.2.54
nvidia-curand-cu12 10.3.2.106
nvidia-cusolver-cu12 11.4.5.107
nvidia-cusparse-cu12 12.1.0.106
nvidia-cusparselt-cu12 0.6.2
nvidia-nccl-cu12 2.19.3
nvidia-nvjitlink-cu12 12.4.127
nvidia-nvtx-cu12 12.1.105
oauthlib 3.2.2
omegaconf 2.3.0
opencv-python 4.11.0.86
packaging 24.2
PasteDeploy 3.1.0
pbkdf2 1.3
pillow 10.2.0
pip 25.0
plaster 1.1.2
plaster-pastedeploy 1.0.1
platformdirs 4.3.6
proglog 0.1.10
propcache 0.3.0
protobuf 5.29.3
psutil 7.0.0
pyav 11.4.1
pycocotools 2.0.8
pycocotools 2.0.8
pydantic 2.10.6
pydantic_core 2.27.2
pyramid 2.0.2
pyramid-mailer 0.15.1
python3-openid 3.2.0
pytorch-lightning 2.5.0.post0
PyYAML 6.0.2
repoze.sendmail 4.4.1
requests 2.32.3
requests-oauthlib 2.0.0
sentry-sdk 2.23.1
setproctitle 1.3.5
setuptools 75.8.0
six 1.17.0
slack_sdk 3.35.0
smmap 5.0.2
SQLAlchemy 2.0.30
sympy 1.13.1
tensorboardX 2.6.2.2
torch 2.2.2
torchaudio 2.2.2
torchmetrics 1.6.3
torchvision 0.17.2
tqdm 4.67.1
transaction 4.0
translationstring 1.4
triton 2.2.0
typing_extensions 4.12.2
urllib3 2.3.0
velruse 1.1.1
venusian 3.1.0
wandb 0.19.8
WebOb 1.8.7
wheel 0.45.1
WTForms 3.1.2
wtforms-recaptcha 0.3.2
yarl 1.18.3
zope.deprecation 5.0
zope.interface 6.4.post2
zope.sqlalchemy 3.1
```

### More info

_No response_

cc @ethanwharris

Start by comparing tests/tests_pytorch/strategies/test_multi_model_ddp.py with tests/tests_pytorch/strategies/test_ddp_integration.py, especially the lines referenced from PR #20936. Reproduce the proposed MultiModelDDPStrategy test with the unused-parameter model and inspect the resulting behavior. Done means the test reliably verifies the intended unused-parameter handling and matches the expected RuntimeError wording.

Written by the indexing model from the issue text.