Sequential offloading bug with Stable Audio

### Describe the bug

Sequential offloading doesn't work when using `pytest`, but does seem to work outside of tests.
This is an issue, because we can't properly test sequential offloading on Stable Audio.

**Additional information:** I've managed to pinpoint where the first `NaN` appears. It's in the `autoencoder` level [when passing through the decoder's `conv1`](https://github.com/ylacombe/diffusers/blob/88933735d0dab2a5b0a565043b1ebdbc0d9f303d/src/diffusers/models/autoencoders/autoencoder_oobleck.py#L283). It might has to do with the fact that we're using [`weight_norm`](https://github.com/ylacombe/diffusers/blob/88933735d0dab2a5b0a565043b1ebdbc0d9f303d/src/diffusers/models/autoencoders/autoencoder_oobleck.py#L264), but at this point, I really don't know since it works outside of the testing file.

### Reproduction

# **Offloading with NaN**

`test_sequential_cpu_offload_forward_pass` and `test_sequential_offload_forward_pass_twice`

# **Working offloading**

```python
import torch
from diffusers import StableAudioPipeline, StableAudioProjectionModel, StableAudioDiTModel, EDMDPMSolverMultistepScheduler, AutoencoderOobleck
from transformers import T5EncoderModel, T5Tokenizer

def get_dummy_components():
torch.manual_seed(0)
transformer = StableAudioDiTModel(
sample_size=32,
in_channels=6,
num_layers=2,
attention_head_dim=4,
num_key_value_attention_heads=2,
out_channels=6,
cross_attention_dim=4,
time_proj_dim=8,
global_states_input_dim=48,
cross_attention_input_dim=24
)
scheduler = EDMDPMSolverMultistepScheduler(
solver_order=2,
prediction_type="v_prediction",
noise_preconditioning_strategy="atan",
sigma_data=1.0,
algorithm_type="sde-dpmsolver++",
sigma_schedule="exponential",
noise_sampling_strategy="brownian_tree",
)
torch.manual_seed(0)
vae = AutoencoderOobleck(
encoder_hidden_size=12,
downsampling_ratios=[1, 2],
decoder_channels=12,
decoder_input_channels=6,
audio_channels=2,
channel_multiples=[2, 4],
sampling_rate=32,
)
torch.manual_seed(0)
t5_repo_id = "hf-internal-testing/tiny-random-T5ForConditionalGeneration"
text_encoder = T5EncoderModel.from_pretrained(t5_repo_id)
tokenizer = T5Tokenizer.from_pretrained(t5_repo_id, truncation=True, model_max_length=25)

torch.manual_seed(0)
projection_model = StableAudioProjectionModel(
text_encoder_dim=text_encoder.config.d_model,
conditioning_dim=24,
min_value=0,
max_value=256,
)
components = {
"transformer": transformer,
"scheduler": scheduler,
"vae": vae,
"text_encoder": text_encoder,
"tokenizer": tokenizer,
"projection_model": projection_model,
}
return components

def get_dummy_inputs():
generator = torch.manual_seed(0)
inputs = {
"prompt": "A hammer hitting a wooden surface",
"generator": generator,
"num_inference_steps": 2,
"guidance_scale": 6.0,
}
return inputs

components = get_dummy_components()
pipeline = StableAudioPipeline(**components)
pipeline.enable_sequential_cpu_offload(device="cuda")

inputs = get_dummy_inputs()
pipeline(**inputs)
```

### Logs

_No response_

### System Info

```
- 🤗 Diffusers version: 0.30.0.dev0
- Platform: Linux-5.4.0-166-generic-x86_64-with-glibc2.29
- Running on Google Colab?: No
- Python version: 3.8.10
- PyTorch version (GPU?): 2.3.1+cu121 (True)
- Flax version (CPU?/GPU?/TPU?): 0.7.2 (cpu)
- Jax version: 0.4.13
- JaxLib version: 0.4.13
- Huggingface_hub version: 0.23.4
- Transformers version: 4.42.0.dev0
- Accelerate version: 0.31.0
- PEFT version: 0.11.1
- Bitsandbytes version: not installed
- Safetensors version: 0.4.3
- xFormers version: not installed
- Accelerator: NVIDIA A100-SXM4-80GB, 81920 MiB
NVIDIA A100-SXM4-80GB, 81920 MiB
NVIDIA A100-SXM4-80GB, 81920 MiB
NVIDIA DGX Display, 4096 MiB
NVIDIA A100-SXM4-80GB, 81920 MiB
- Using GPU in script?: yes
- Using distributed or parallel set-up in script?: no
```

### Who can help?

cc @yiyixuxu @sayakpaul

Start by running `test_sequential_cpu_offload_forward_pass` and `test_sequential_offload_forward_pass_twice`, then inspect `src/diffusers/models/autoencoders/autoencoder_oobleck.py` around the decoder's `conv1` and `weight_norm`. Compare the failing pytest path with the provided standalone Stable Audio reproduction; done means sequential offloading no longer produces NaNs in either test.

Escrito por el modelo de indexación a partir del texto del issue.