Improve test coverage and fix map enumeration bug in DAG task filters

[Copilot]

Test coverage was minimal, leaving core DAG, task, run, and store logic largely untested. Three functions also had a silent runtime bug due to incorrect map enumeration.

Bug Fix

get_completed_tasks/1, get_pending_tasks/1, and get_running_tasks/1 iterated over dag.tasks (a map) but passed the resulting {key, value} tuples directly to DAGTask.is_completed/1 etc., which pattern-match on %DAGTask{} — causing FunctionClauseError on any non-empty call.

# Before — crashes at runtime when tasks exist
Enum.filter(dag.tasks, fn task -> DAGTask.is_completed(task) end)

# After — correctly destructures map entries
Enum.filter(dag.tasks, fn {_task_id, task} -> DAGTask.is_completed(task) end)

New Test Files

• test/dag_task_test.exs — DAGTask.new/1, validate/1, is_pending/1, is_completed/1, is_running/1, status helpers, set_handler/2
• test/dag_run_test.exs — DAGRun.new/1, generate_id/0, random_string/1, uniqueness guarantees, timestamp fields
• test/dag_task_run_test.exs — DAGTaskRun.new/3, field initialization, payload storage, timestamps

Expanded Coverage in ex_dag_test.exs

Additional ExDag.DAG cases covering: get_tasks/1, set_handler/2, set_default_task_handler/2, set_tasks_handler/2, validate_for_run/1 (disconnected graph), get_deps/2, get_deps_map/1, get_runs/2, status accessors, completed?/1, get_completed_tasks/1, get_pending_tasks/1, get_running_tasks/1, sorted_tasks/1, should_run_task/2, clear_failed_tasks_runs/1, String.Chars impl, keyword-based add_task, and add_task!/2 throw behavior.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.