fix(distributed): execute work units and track task outcomes

[adolago]

Summary

• implement real work-unit task execution in Controller::execute_work_unit
• execute supported task modules (command, shell) using task module + args
• persist per-task results (stdout/stderr/exit code), errors, and final work-unit state
• mark work units Failed on task failures instead of placeholder completion
• add regression tests for non-placeholder completion behavior and failure-state tracking

Closes #787

Test evidence

• cargo test --features distributed distributed::controller::tests -- --nocapture
• Result: 4 passed; 0 failed
• Covered tests:
  • test_controller_creation
  • test_controller_info
  • test_execute_work_unit_not_completed_when_task_cannot_execute
  • test_execute_work_unit_marks_failed_on_task_failure

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ee6ed250eb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Honor when conditions before running tasks

execute_work_unit now executes every task in order without checking TaskSpec.when, so any conditional task is run unconditionally. In workflows that submit work units with when expressions (for example, guard tasks that should be skipped on certain hosts/variables), this can trigger unintended commands and false failures instead of a skip, which changes playbook semantics and can produce incorrect side effects.

Useful? React with 👍 / 👎.