Add checks to assess if the server machines are correctly pinging the API

[AlbertoCasasOrtiz]

We can do this in the dequeue function, save the information about last time a machine pinged the API in a local file (in the API machine, be careful with race conditions), and send an email in case a machine has not been pinging the API for more than a day (discuss period of time).

[carmichaelong]

Some other possible ideas:

• For each IP, provide the latest associated trial "updated at" time. When servers are busy enough, will be easy to see that there is likely an issue that happened a few hours back.
• For each IP, maybe the last ~5-10 status (error or done) counts to get info on if a machine might be starting to fail. There might be some misleading signals, though, if someone doesn't have the right setup and has a lot of errors for calibration or neutral trials.