Change the 4C CLI parser to CLI11

[maxiludwig]

This pull request modernizes and streamlines the command-line argument parsing and I/O handling in the 4C global_full application. The main improvements are the introduction of the CLI11 library for argument parsing, the removal of legacy parsing logic, and the refactoring of input/output handling to support more robust and flexible usage, especially for nested parallelism.

Key changes include:

• Added CLI11 as a required dependency to enable modern command-line parsing.
• Introduced helper functions in 4C_global_full_io.cpp to adapt old legacy style command line arguments to make them parsable by CLI11. The sigle dashed options (-ngroup, -glayout, etc) and the restart arguments (restart= and restartfrom=) are internally converted to double dashed option style --option. Warnings are thrown if the old option style is used.
• Most of the validity checks can now run when parsing the CLI input in 4C_global_full_main.cpp. Some checks doing cross argument validation are done right after that.
• Updated the CommandlineArguments struct to use the already parsed arguments. Downstream functionalities (like 4C_comm_utils.cpp) are no longer doing the actual parsing.
• Refactored the logic for determining input/output/restart file identifiers into a single update_io_identifiers function, improving clarity and reducing code duplication.

[sebproell]

Thanks @maxiludwig! I left a first review. In summary, the code requires some restructuring to facilitate testing.

[sebproell]

Do we need this? I guess this is just for the help message. If there are no serious flaws with the defaults (and I doubt there are any), I would just use what CLI11 provides.

[maxiludwig]

POSITIONALS:
  input TEXT                  Name of the input file, including the suffix
  output TEXT                 Prefix of your output files.
  io_pairs TEXT ...           More pairs of simulation <input> and <output> names. Only
                              necessary when using nested parallelism with multiple groups and
                              separate input files.

I found this a bot too dense, so i thought having the blank lines makes reading of the options easier.

 POSITIONALS:
  input TEXT                  Name of the input file, including the suffix

  output TEXT                 Prefix of your output files.

  io_pairs TEXT ...           More pairs of simulation <input> and <output> names. Only
                              necessary when using nested parallelism with multiple groups and
                              separate input files.

[sebproell]

I think the other version is also fine but I agree that the spaced version is slightly more readable. By the way, could you post the full help message in the description of the PR?

[Copilot]

Pull Request Overview

This pull request modernizes the command-line argument parsing in 4C by replacing legacy custom parsing logic with the CLI11 library. The changes introduce more robust argument validation, better error messages, and support for both legacy (single-dash) and modern (double-dash) option formats with deprecation warnings.

Key changes:

• Integration of CLI11 library as a required dependency for modern command-line parsing
• Introduction of helper functions to adapt legacy command-line arguments (single-dash options like -ngroup and dashless options like restart=) to CLI11-compatible format with deprecation warnings
• Refactoring of I/O identifier logic into a centralized update_io_identifiers function

Reviewed Changes

Copilot reviewed 22 out of 22 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
cmake/configure/configure_CLI11.cmake	Configures CLI11 as a fetchcontent dependency (version 2.6.1)
CMakeLists.txt	Adds CLI11 as a required dependency
cmake/setup_install.cmake	Adds CLI11 to the list of dependencies in the config file
cmake/templates/4CConfig.cmake.in	Adds CLI11 package finding to the config template
src/core/comm/src/4C_comm_utils.hpp	Adds CommConfig struct and updates function signature to accept config pointer instead of argv vector
src/core/comm/src/4C_comm_utils.cpp	Refactors create_comm to use CommConfig struct instead of parsing argv internally
src/core/comm/tests/4C_comm_utils_test.np3.cpp	Updates test to use new CommConfig-based API
src/core/io/src/4C_io_command_line_helpers.hpp	Defines CommandlineArguments struct and helper functions for argument parsing and validation
src/core/io/src/4C_io_command_line_helpers.cpp	Implements legacy CLI adaptation, I/O pair building, cross-compatibility validation, and I/O identifier updates
src/core/io/tests/4C_io_command_line_helpers_test.cpp	Adds comprehensive unit tests for command-line helper functions
apps/global_full/4C_global_full_main.cpp	Replaces custom help text and parsing with CLI11-based parse_command_line function
apps/global_full/4C_global_full_io.hpp	Updates function signatures to accept CommandlineArguments struct and separate communicators
apps/global_full/4C_global_full_io.cpp	Removes legacy parsing functions and updates setup functions to use new argument structure
unittests/mat/4C_electrode_test.cpp	Updates to use new CommConfig-based communicator creation API
tests/input_files/*.4C.yaml	Updates documentation in test input files to show double-dash option format
doc/documentation/src/**/*.rst	Updates documentation examples to use double-dash option format
cmake/functions/four_c_testing_functions.cmake	Updates test command generation to use double-dash format

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[sebproell]

@maxiludwig I think this is a nice improvement once you address the comments. Especially the nested parallelism looks like it could need more rework, but I like the increment and think we can merge.

[sebproell]

Nice, the struct is much easier to understand than the old vector of strings 👍 Please document all these fields and their interaction here. Isn't n_group just group_layout.size()? If yes, drop it.

[georghammerl]

When everything is setup properly, this needs to be the case.

[sebproell]

Suggested change

	Communicators create_comm(CommConfig& config);
	Communicators create_comm(const CommConfig& config);

[sebproell]

This is placed wrongly now. The concrete parameters are a choice of the 4C main executable. Think about it this way: other executables should be able to use general CLI utilities that are in core, but of course they should not add a struct listing their parameters. Please move the struct back to the apps. There are a few functions that perform work on the struct, so they likely also belong to apps. Only the logical manipulations like the single-dash/no-dash fixes can be part of core. And any other helpers that we like to interface with CLI11. Currently, you use CLI11 directly in apps and I think that is fine. If we ever build abstractions on top of CLI11, they would again go in core. I am writing this to hopefully paint a clearer picture of how to sort functionality into the correct modules. I am not saying you should implement any kind of general cli interface right now.

[sebproell]

I think the other version is also fine but I agree that the spaced version is slightly more readable. By the way, could you post the full help message in the description of the PR?

[georghammerl]

Suggested change

	"'diffgroupx' can be used to compare results from separate but parallel 4C runs; "
	"'diffgroupx' can be used to compare vectors/matrices/results between two separate (serial/parallel) 4C runs; "