Prout ascribes Testing-in-Production results to a PR before the PR is even seen in Prod

In membership frontend, 2 PRs were merged within a few minutes of each other:

1. [PR 1436](https://github.com/guardian/membership-frontend/pull/1436)
2. [PR 1441](https://github.com/guardian/membership-frontend/pull/1441)

Although PR 1436 went out first, _all_ the Testing-In-Production results were posted by Prout onto PR 1441. Though it can be hard to accurately ascribe a test to PR, what went **particularity** wrong here was that the TiP results [were ascribed](https://github.com/guardian/membership-frontend/pull/1441#issuecomment-273739485) to PR 1441 before it was even _seen_ in Production:


Start by comparing the deployment and production-visibility timing for PR 1436 and PR 1441, then inspect the TiP result attribution path. Reproduce the case where results are attached to PR 1441 before it is seen in production; done means the results are attributed to the appropriate PR only after the relevant production evidence is available.

Written by the indexing model from the issue text.