Hide `CancellationToken` parameters from CLI help output

[lex57ukr]

Goal

Ensure keystone-cli --help output does not expose CancellationToken as a user-facing argument (e.g., --cancellation-token) for any command.

Scope

• Audit all Cocona command implementations and identify any methods that accept CancellationToken in the signature and therefore surface it in help output
• Determine the preferred approach:
  • remove CancellationToken from command method signatures where it’s not needed, or
  • suppress/hide it from Cocona-generated help output (if Cocona supports this)
• Apply the chosen approach consistently across all commands
• Add a unit test that scans command methods via reflection to enforce the rule:
  • CancellationToken is not present in command method parameters unless it is explicitly suppressed from help output

Outcome

• CLI help output contains only user-actionable arguments
• No command lists --cancellation-token (or similar) in help text
• A regression test prevents this from reappearing in future commands

Notes

• This is a UX/ergonomics improvement, not a behavior change.
• If Cocona doesn’t support suppression, standardize on “don’t include CancellationToken in command signatures” and document the convention.

[lex57ukr]

Triage

Guidance on making the test “rock solid” (without being brittle)

Instead of parsing full --help output (which can be formatting-sensitive), prefer this order:

1. Rule-based reflection test (stable, fast)

   • Discover command handler methods from their registration
   • Assert no parameters of type CancellationToken
   • If suppression is supported, allow it only when a clear suppression marker is present (attribute or convention)

2. Help text snapshot test (only if needed; more brittle)

   • Generate Cocona help output in-memory
   • Assert it does not contain cancellation-token
   • Useful as an additional guard, but not required if reflection coverage is sufficient

Optional: split into two issues for smaller PRs

If you want to keep PRs very tight:

1. Audit and decide on the approach, then implement suppression/removal
2. Add a reflection-based regression test and document the convention

It’s also reasonable to address this in a single issue/PR.