I/Q: Reduce the verbosity of the output of `get_command_info`

Even with the largest models, one quickly runs out of context when using I/Q due to the output of standard tools like `get_command_info` being too detailed by default.

**Task:**
Study how the default output of `get_command_info` could be reduced and more control given to the LLM regarding what information should be queried.

Start by locating the I/Q implementation of get_command_info and inspect how its default output is assembled. Study which details can be omitted by default and how the LLM could request specific information. Done means the standard output uses less context while allowing more selective queries.

Written by the indexing model from the issue text.