SIGSEGV in cs_sles_solve_ccc_fv when CS_CUDA_ALLOC_DEVICE_UVM is not set (v9.1.0)

## Description

`cs_sles_solve_ccc_fv` allocates its extended solver buffers using `CS_ALLOC_DEVICE` (device-only memory). Those buffers are subsequently passed to the CPU-side convergence check, which dereferences them on the host and causes a SIGSEGV.

Setting `CS_CUDA_ALLOC_DEVICE_UVM=1` before launching the solver works around the crash: it remaps `cs_alloc_mode_device` to `CS_ALLOC_HOST_DEVICE_SHARED` at initialisation time (see `cs_base_cuda.cu:777`), making the buffers accessible from both host and device.

## Steps to reproduce

1. Build code_saturne v9.1.0 with CUDA support.
2. Run any case that exercises the CUDA sparse linear solver (`cs_sles_solve_ccc_fv`) without setting `CS_CUDA_ALLOC_DEVICE_UVM`.
3. Observe SIGSEGV during the convergence check after the first solver call.

## Workaround

```bash
export CS_CUDA_ALLOC_DEVICE_UVM=1
```

Set this before invoking the solver. It prevents the crash by ensuring solver buffers are in unified/shared memory.

## Expected behaviour

`cs_sles_solve_ccc_fv` should either allocate convergence-check buffers in host-accessible memory directly, or the convergence check should use a device-to-host copy rather than a direct host dereference.

## Environment

- code_saturne version: 9.1.0
- CUDA version: 13.1
- GPU architecture: sm_75
- OS: Ubuntu (x86_64)

Commencez par cs_sles_solve_ccc_fv et examinez comment ses tampons étendus du solver parviennent à la vérification de convergence côté CPU. Comparez le comportement d’allocation avec cs_base_cuda.cu:777 et reproduisez une compilation CUDA sans CS_CUDA_ALLOC_DEVICE_UVM. C’est terminé lorsque le solver ne provoque plus de segmentation fault sans le workaround, tandis que le workaround Unified Memory existant reste valide.

Rédigé par le modèle d'indexation à partir du texte de l'issue.