Restrict product extraction to allowlisted domains

[Osmose]

@javaun @elancaster Given that we're now looking to ship via Test Pilot, do we still want to restrict the domains we attempt product extraction on? If not, we risk more incorrect extraction, but we give users the opportunity to report failure on domains that aren't in our original 5. We could also temporarily restrict for launch and open up after we have a better idea about how to handle ongoing development of our Fathom ruleset.

[javaun]

When we start collecting opt-in data, one compelling reason to have it just run without a whitelist is that we'd collect data differently for shopping pages vs. other sites:

1. If it's not a recognized shopping product page, just send the domain name. This is one way we'll discover which long-tail sites users shop.
2. If Fathom does recognize a product page (whether it's an officially supported one, or Etsy, which isnt' supported but sometimes works), send the full URL.

That said, until we are ready to start collecting data, I think we could still restrict Fathom extraction to whitelisted domains. Fathom is new tech and there could easily be perf issues. If so, isolating it to just 5 sites may make it easier to troubleshoot. What do you think?

[Osmose]

Sounds like "initially restrict, and revisit post-launch" is the way to go.