0.5.1 - Better handling of transient network failures

[shawnrushefsky]

• Adds clearer reasons when invoking the imds reallocate endpoint
• We also now handle the case where the heartbeat manager throws an error. This is typically because of a transient inability to access the kelpie api. This could happen because a router got unplugged, the internet went out on a node, or there was an error effecting the kelpie api or its dependent services on cloudflare. Now, we will keep trying to restart the heartbeat until the node dies or connection is restored.
• But what if my job already got handed out again during disconnection? The heartbeat endpoint will return a status of "canceled" if a heartbeat is attempted against a job that is held by another machine id. This will make the kelpie worker treat the job as a normal canceled job, interrupting the job process and asking for another job.

[github-actions]

🚀 Download the latest release candidate 🚀

[rxsalad]

Let's go!