[ETCM-829] rewrite healthcheck

[AurelienRichez]

Description

This PR rewrite the healthcheck endpoint to have more precise info :

• it checks that we have at least one peer (before it only checked that the service responded without an error, but not the content of the response itself)
• it returns block numbers for best stored block, best known block and best block we are fetching
• it returns a 500 when the state did not change change for too long
• it returns the sync status (STARTING, SYNCING, SYNCED)

It can still be improved by returning the number of healthy peers instead of the raw number of peers.

Important Changes Introduced

The healthcheck endpoint is not backward compatible (for instance I renamed the field description to name because it seems more accurate)

two new parameters :

• mantis.network.rpc.http.health.no-update-duration-threshold
• mantis.network.rpc.http.health.syncing-status-threshold

The values of some parameters are given as a generic parameter "info" which is a string. I think it is alright for a healthcheck endpoint but we might want to consider if we want something more strongly typed.

Testing

You can call <node_url>/healthcheck to see the endpoint, and check that it moves correctly

The result should look like this :

{
  "checks": [
    {
      "name": "peerCount",
      "status": "OK",
      "info": "3"
    },
    {
      "name": "bestStoredBlock",
      "status": "OK",
      "info": "80000"
    },
    {
      "name": "bestKnownBlock",
      "status": "OK",
      "info": "12721559"
    },
    {
      "name": "bestFetchingBlock",
      "status": "OK",
      "info": "80000"
    },
    {
      "name": "updateStatus",
      "status": "OK"
    },
    {
      "name": "syncStatus",
      "status": "OK",
      "info": "SYNCING"
    }
  ]
}

[dzajkowski]

looks good, a few small comments