fix(robot-server): Publish to maintenance_runs/current_run on initial PLAY action

[mjhuff]

Overview

When a client posts a PLAY action that starts a protocol run, any existing maintenance run ID is cleared from the server. This means that MQTT should update subscribers interested in the current maintenance run.

I've decided to follow the existing pattern for how we've handled async/sync publishing in the past, which is create a sync version if one doesn't exist instead of changing the method that uses the publish to async.

Test Plan and Hands on Testing

• Verified via Postman that the refetch at robot-server/maintenance_runs/current_run occurs when playing a protocol run for the first time that run.

Changelog

• Fixed starting protocol runs not properly alerting the desktop app and ODD that maintenance runs should be closed.

Risk assessment

low

[sfoster1]

Okay, nice!

[mjhuff]

Apologies for the quick merge, need this for another PR going up in a few. Will address feedback in a separate PR, if necessary.