Invesitgate the best way to provide metadata context for why some Execution has failed

[anna-geller]

We were investigating adding a function errorInfo() that would emit the following as a JSON object:

{
  "namespace": "namespace",
  "flowId": "flowId",
  "id": "id",
  "startTime": "...",
  "endTime": "...",
  "tasks": [{
    "id": "id",
    "type": "task class like Python, SQL Server Query etc"
  }],
  "state": "FAILED"
}

If the function were called from a non-FAILED execution, the tasks array would be null.

However, there were some challenges with this approach:

• The execution is not terminated when you are in an error branch, so it is reported as RUNNING. This means we cannot limit the function to FAILED execution as we have no way to know if an execution is FAILED before it is terminated.
• We can fetch an execution that is not up-to-date, so the function can return no failed task even if some failed task run exists

So the errorInfo() as a function doesn't work.

This will solve the problem raised here https://github.com/kestra-io/kestra-ee/issues/2264#issuecomment-2500397364