Enhance CLI UX by supporting shell expansion for claude-flow sub-options (nested subcommand autocompletion)

Description:

Currently, the claude-flow CLI tool does not support shell autocompletion for its nested subcommands or sub-options. This degrades user experience of someone staying at the Linux command line, particularly when dealing with multi-level command trees that require memorization or frequent consultation of --help.

This feature request is to implement shell expansion (autocomplete) support for claude-flow, inspired by techniques described in this Stack Overflow discussion:
🔗 https://stackoverflow.com/questions/17879322/how-do-i-autocomplete-nested-multi-level-subcommands

Use Case
Example of desired UX improvement:

```
$ claude-flow
analyze batch config help

$ claude-flow config
set get reset

$ claude-flow config set
api-key model timeout

```
This would support nested structures with arbitrary depth, improving discoverability and productivity and bring shell-level usability to more similar to being within the claude shell.

Suggested Implementation Approaches
- Leverage argcomplete or similar Python tools that support dynamic completions.
- Generate a _claude-flow shell completion script with support for nested subcommands, as shown in the linked Stack Overflow example.

Benefits

- Faster command-line usage
- Lower cognitive load for new users
- Easier scripting and automation

Related
- [How do I autocomplete nested, multi-level subcommands?](https://stackoverflow.com/questions/17879322/how-do-i-autocomplete-nested-multi-level-subcommands) — helpful discussion on shell expansion design patterns.

Would love to see this enhancement prioritized for smoother CLI interactions!

Start by locating the claude-flow CLI entry point and its nested command definitions. Review the linked nested-subcommand completion discussion and assess argcomplete or generated shell completion scripts. Done means completion exposes top-level, nested, and deeper sub-options such as the examples in the issue.

Written by the indexing model from the issue text.