CircleCI test to validate from-bodyxml

[adgad]

Problem

We need a way to validate that our migration library (from-bodyxml) works for all of FT pink's content

Proposed solution

A test in CircleCI that:

• gets CAPI notifications from the last 5 minutes
• runs from-bodyxml on them
• validates against the generated JSON schemas for ContentTree.transit

Example: https://app.circleci.com/pipelines/github/Financial-Times/content-tree/453/workflows/98438d49-c044-48d6-98c7-6ae43fcc3593/jobs/445/steps

Limitations

• This is useful for recent articles, but isn't likely to pick up old/archived content (where there is likely to be weirder stuff)
• It will validate that the tree we produce is valid, but not necessarily correct (for example, if our transformer chose an incorrect but still valid value for layoutWidth, this test would pass). A possible improvement here is to also compare the output against what is being created by cp-content-pipeline-api (assuming that is a reasonable source-of-truth for correctness, given it currently serves the website).

Despite the limitations, hopefully this is still useful at least for an indication of how close to complete the migration code is.

[epavlova]

Despite the limitations mentioned, I think it's useful.