Cache failed requests to WordPress.org API endpoints

[peterwilsoncc]

WordPress does not cache failed responses to various WordPress.org API endpoints (theme & plugin updates, etc).

When disconnected from a network this significantly impacts the performance of the dashboard as the requests are repeated as the user navigates:

• Main dashboard page: 2 requests leading to 20 second delay
• Ajax request to planet WP: 10 seconds delay, non blocking
• Plugin dashboard page: 2 requests to the same endpoint, 20 second delay
• Theme dashboard page: 2 requests to the same endpoint, 20 second delay
• Updates dashboard page: 6 requests (3 repeated), 60 second delay
• Site health dashboard page: 1 blocking request, 10 second delay
• Site health dotorg-communication endpoint: 20 second delay
• Site health debug page: 7 blocking requests, 70 second delay

Successful requests to most endpoints are cached for twelve hours, I think it would be wise to cache unsuccessful requests for some amount of time but probably a little less. 6 hours maybe???

For the site health screens, it may be wise to force uncached requests so up-to-date information is known but this ought to occur via the site's REST API so the requests are non blocking.

I picked this up while testing ​PR#7350.

To reproduce:

1. Configure WordPress locally.
2. Disconnect your computer from the network
3. Run the WP CLI commands: wp transient delete --all; wp transient delete --all --network; wp cache flush
4. Navigate around the WordPress admin