Idea from @Krinkle at the offsite: Deploy a static version of an article on our domain so we can have it as a benchmark to "know" better if the problem is on Wikipedia or the tool we are using. We tried out having only the header/footer before but that is so fast so it didn't help us, we need to have some actual content on that reference page.
|performance/WebPageTest : master||Test a static test page (both in FF and Chrome)|
|operations/mediawiki-config : master||speed-tests: Actually add the images|
|operations/mediawiki-config : master||speed-tests: Make static rev 872156204 more standalone|
|operations/mediawiki-config : master||speed-tests: Remove old content and add a new article copy|
The static page now exists. Once monitored by WebPageTest/WebPageReplay, it will allow us to differentiate betweenp perf impact triggered by wiki-related changes (MediaWiki software, its configuration, and user-generated content) and infrastructure changes (e.g. Varnish/Nginx in prod, or the proxy and browser we use and the test server it runs on).
I've filed T211487 for the similar task for an on-wiki page.
Added it to both WebPageTest and WebPageReplay, they are under static: