improve command definition

[eviltester]

merge command definitions and help to make it easier to maintain

remove the faker-command-help-metadata and rely on the domain

reduce faker-command-help-metadata.js

rely on domain-faker-module-keyword-definitions.js

[eviltester]

Issue 228 is only partially covered by the current PR. The current branch has moved a lot of help metadata into the keyword definition files, but the old compatibility layer is still present and consumers still depend on it.

Remaining work to fully complete this issue:

1. Remove the remaining faker command help compatibility layer as a source of truth.

   • packages/core/js/faker/faker-helper-keyword-definitions.js should stop adapting domain help back into a faker-command-help shape for domain commands.
   • Keep helpers.* definitions there if needed, but do not let that file remain the main help lookup path for domain commands.

2. Make domain definitions the direct source for domain command help consumers.

   • Update UI/help consumers that currently call getFakerCommandHelp(...) for domain commands to use domain help directly.
   • This includes places like the help model builder and matrix/support code that currently import @anywaydata/core/faker/faker-helper-keyword-definitions.js.

3. Split the concept of "domain command help" from "faker helper help" in consuming code.

   • Domain commands should resolve from domain keyword definitions.
   • helpers.* commands should resolve from faker helper definitions.
   • Consumers that currently use one mixed lookup API should be refactored to choose the correct source explicitly.

4. Reduce or remove the legacy faker-command-help naming in tests and consumers.

   • Rename tests and helpers that still talk about faker-command-help-metadata when they are now really validating domain keyword help or helper keyword help.
   • The code should describe the real architecture after the refactor.

5. Add explicit regression tests for the new ownership boundaries.

   • Assert that domain command metadata is resolved from domain definitions.
   • Assert that helpers.* metadata is resolved from faker helper definitions.
   • Assert that no consumer needs the old mixed compatibility mapping for domain commands.

6. Only close this issue when the compatibility adapter is no longer needed for normal domain help resolution.

   • At that point the definitions and help model will genuinely be merged in the maintainable way this issue asks for.

So in short: this PR provides a lot of the groundwork, but issue 228 is not fully done until consumers stop depending on the old mixed faker-help compatibility path for domain commands.