Use different ports for different GPU IDs

[allenwang28]

This will enable multiple vLLM replicas to be spun up on the same local host.

Tested this by changing

services:
  policy:
    procs: 2
    num_replicas: 1
    with_gpus: true

in apps/vllm/llama3_8b.yaml.

Doing so without my change showed this:

  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/vllm/v1/worker/gpu_worker.py", line 578, in init_worker_distributed_environment
    init_distributed_environment(parallel_config.world_size, rank,
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/vllm/distributed/parallel_state.py", line 976, in init_distributed_environment
    torch.distributed.init_process_group(
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper
    return func(*args, **kwargs)
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 95, in wrapper
    func_return = func(*args, **kwargs)
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1752, in init_process_group
    store, rank, world_size = next(rendezvous_iterator)
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/torch/distributed/rendezvous.py", line 230, in _tcp_rendezvous_handler
    store = _create_c10d_store(
  File "/home/allencwang/.conda/envs/forge_test_2/lib/python3.10/site-packages/torch/distributed/rendezvous.py", line 198, in _create_c10d_store
    return TCPStore(
torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 12345, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use