Initial AP poller implementation

[bryophyta]

What does this change?

Adds a new poller lambda, using the 'long polling' pattern added in #83, to get data from the AP /feed endpoint and pass it to the Newswires ingestion lambda. Getting the data needed for the ingestion lambda requires two steps:

1. Long-polling request to the /feed endpoint, which returns immediately if there are new items since the last request (determined by the next_page value in the response from the previous request, and passed forward to the lambda via the SQS queue that invokes it) or times out after 15 seconds. (See AP docs)
2. The /feed endpoint returns only part of the data we're interested in. To get the 'body' of the story we need to fetch and parse the NITF representation of the story, which is in an XML format.

nb. we don't yet have a clear standard for judging when our processing of a feed is complete/production-ready, so it's possible that we'll want to make some changes later (e.g. if we discover we're missing important fields).

How to test

• Run locally, using npm run dev -w poller-lambdas and selecting the AP poller.
• Deploy to CODE; check that it queues up the next invocation with next_page as expected.
• Run the unit tests which check that we're extracting the expected values from some samples of the NITF rendition of a couple of stories.

How can we measure success?

Have we considered potential risks?

Images

Accessibility

• Tested with screen reader
• Navigable with keyboard
• Colour contrast passed
• The change doesn't use only colour to convey meaning

[github-actions]

Deploy build 505 of Editorial Tools::News Wires to CODE

All deployment options

• Deploy build 505 of Editorial Tools::News Wires to CODE
• Deploy parts of build 505 to CODE by previewing it first
• What's on CODE right now?

---

From guardian/actions-riff-raff.