Master Plans CLI ergonomics and command wrappers

Problem

The tasks task management system requires invoking raw Python commands (uv run python tasks/plans_cli.py task create ...) for every operation, making it cumbersome for both human operators and AI agents. There is no registered CLI entrypoint, no shorthand aliases for common operations (start/complete a task), and no discovery commands (list tasks by status, find by GitHub issue). This friction discourages consistent task tracking and makes the cbox manager workflow unnecessarily verbose — every task operation requires remembering the full module path and argument syntax.

Context

Possible Solutions

Plan

Implementation Progress

• Add a friendly callable entrypoint (for example plans) so usage does not require uv run python ....
• Add command shortcuts/wrappers for common flows (task create, task update, task start, task complete).

• Add discovery/status commands (for example task list, task show, task find --github-issue).

• A user can create/update/find tasks with concise commands and no raw python invocation.

• Existing plans_cli.py behavior remains compatible.

• New commands have focused tests (or equivalent validation) and docs examples.

• User request driver: improve ergonomics for master plan execution and cbox-manager usage.

• Implemented command surface:
• plans task start --path ... (alias for status in_progress)
• plans task complete --path ... (alias for status completed)
• plans task list --workstream <ws> [--status ...]
• plans task show --path ...
• plans task find --github-issue <id>
• Entrypoint plans registered in pyproject.toml [project.scripts].
• CLI reference docs at tasks/tools/cli-reference.md.
• 12 focused tests in tests/core/test_plans_cli.py.
• No cross-workstream dependencies; no blockers.
• Follow-up: consider migrating from argparse to typer for consistency with dft CLI (owner: head-of-engineering).

Review Feedback

• [ ] Review cleared