Use back-off algorithm to perform health check on function instance

[thomasgouveia]

What needs to be done ?

We should consider using an implementation of the back-off algorithm in order to determine the readiness of a function instance.

See : https://github.com/cenkalti/backoff

Why it needs to be done ?

Depending on the latency of the network and the function image size (e.g: Node, Python), our current naive implementation of the health check can exit without totally waiting the instance. This is an issue, as the CLI will receive a blank response from the back end.