[SYCL][CUDA] Nvidia GPUs detected under multiple platform

[TejaX-Alaghari]

Describe the bug
On a system with 2 Nvidia GPUs, SYCL detected the 2 devices as being under 2 different platforms. Whereas the expected behavior is for them to be listed under a single Nvidia platform.

To Reproduce

1. A simple program to list the devices in the environment sycl_info.cpp

2. Compilation command: clang++ -fsycl -fsycl-targets=nvptx64-nvidia-cuda sycl_info.cpp -o sycl_info

3. Launch command: ./sycl_info
   Output: SYCLInfo.txt

4. The two devices in the above log are listed under 2 different platforms instead of being grouped under a single platform as shown by clinfo -l
   Output: CLInfo.txt

Environment:

• OS: Linux
• Target device and vendor: Nvidia GPU
• DPC++ version: 16.0.0 (https://github.com/intel/llvm.git 7e37406)
• Dependencies:
  Nvidia Driver Version: 525.60.13
  CUDA Version: 12.0

Additional Information

Log of sycl-ls --verbose sycl_ls.txt
Log of nvidia-smi NVInfo.txt
Log of nvidia-smi -a nvidia_smi_a.txt

[pengtu]

#6446

[JackAKirk]

Closed by #13616