[Task]: Simplify AMO's __heartbeat__ endpoints

[diox]

Description

We have 3 separate kinds of healthchecks and heartbeat endpoints in addons-server:

• External healthcheck uses https://addons.mozilla.org/services/monitor.json which checks a bunch of monitors
• Internal __heartbeat__ endpoint that is used for kubernetes to check a pod is ready
• Internal __lbheartbeat__ endpoint that is used for kubernetes to check a pod is healthy

That last one is just implemented with a middleware that just returns 200, it's fairly minimal. But the middle one, that is still used to check for readiness when spawning pods, makes a call to the database, to elasticsearch, tries to use PIL... that's a lot of CPU wasted, and a lot of connections to the outside world. When elasticsearch or the database are not healthy, it's pointless to consider the pods not ready and try to spawn new ones...

We should look into implementing __heartbeat__ just like __lbheartbeat__.

Acceptance Criteria

• item 1

┆Issue is synchronized with this Jira Task

[diox]

Similar problem on addons-frontend:

• __lbheartbeat__ is implemented in a similar fashion and just immediately returns a 200
• __heartbeat__ calls addons-server site API endpoint, which is a bit aggressive.

[diox]

Implementing this will make addons-server pods less susceptible to restarting when deploying under load.