core: health check should be KO if invalid infra or non-existing (404 from editoast)

[bougue-pe]

What happened?

Core stays "healthy" no matter the outcome of infra loading.
(no health endpoint exists)

This leads to some cases where a timeout from editoast locks core workers and the queue stays full

What did you expect to happen?

We could either add a liveness server or kill the process in case something is going unexpected when loading the infra.
What's expected is:

• loading OK : obvious case
• loading leads to an invalid infra : in this case, stay alive and reject any request on the same infra and same version
• requesting an infra that doesn't exist anymore : stay alive and reject any request on that infra
• any other case should lead to a retry of loading, then shutdown

How can we reproduce it (as minimally and precisely as possible)?

See above

On which environments the bug occurs?

Dev (SNCF)

On which browser the bug occurs?

Firefox

OSRD version (top right corner Account button > Informations)

dev