Device mesh initialization fails when running FSDP on CPU

Nobody has claimed this yet.

### Bug description

When running FSDP on CPU (e.g., for benchmarking or testing), the strategy attempts to initialize the `DeviceMesh` using a hardcoded `"cuda"` device type. This causes a failure on CPU-only environments where CUDA is not available, or when FSDP is intentionally run on CPU.

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

v2.5

### Reproduced in studio

_No response_

### How to reproduce the bug

```python
import torch
from lightning.fabric import Fabric
from lightning.fabric.strategies import FSDPStrategy

# Configure FSDP with device_mesh on CPU
strategy = FSDPStrategy(parallel_devices=[torch.device("cpu")], device_mesh=(1,))
fabric = Fabric(strategy=strategy, devices=1)
fabric.launch()
```

### Error messages and logs

```
File "src/lightning/fabric/strategies/fsdp.py", line 271, in setup_environment
self._fsdp_kwargs["device_mesh"] = init_device_mesh("cuda", self._fsdp_kwargs["device_mesh"])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
...
RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
```

### Environment

Current environment

```
- PyTorch Lightning Version: master/latest
- PyTorch Version: 2.2+ (where DeviceMesh is supported in FSDPStrategy)
- OS: Linux/macOS/Windows
- Accelerator: CPU
```

### More info

The hardcoded `"cuda"` string in `setup_environment` should be replaced with `self.root_device.type` to dynamically support the correct device type.

cc @ethanwharris

Start in src/lightning/fabric/strategies/fsdp.py at setup_environment and reproduce the CPU configuration shown in the issue. Verify that DeviceMesh initialization uses the selected device type and that the CPU run completes without requiring CUDA.

Written by the indexing model from the issue text.