[Profiler] Output grouping information && process and hardware mapping

[phoenixdong]

Add print_ranks method for output during initialization, output the following information:

• parallelism_to_groups.json
• rank_to_host_and_device.json
• rank_to_parallelism_to_group_id.json

Enable this feature by adding the following fields to the configured yaml file, for example:

system:
  ...
  analyze:
    analyze_save_dir: ./examples/aquila/analyze

If not configured, the above file information will be output to stdout by default.