frontend/node-red-dashboard: unknown camera type after restarting Node-RED

[ethanjli]

Describe the bug
After restarting the Node-RED dashboard without also restarting the backend hardware controller, the Node-RED dashboard no longer knows what type of camera is in the PlanktoScope.

To Reproduce
Steps to reproduce the behavior:

1. Go to http://home.pkscope/admin/cockpit/system/services#/nodered.service and log in.
2. In the menu icon (which looks like a vertical ellipsis), restart the service.
3. Go to the "System Monitoring" page of the Node-RED dashboard. Confirm that the "Camera Name" field is empty.
4. On the "Administration" page of the Node-RED dashboard, restart the hardware controller.
5. Return to the "System Monitoring" page of the Node-RED dashboard. Confirm that the "Camera Name" field is no longer empty.

Expected behavior
After restarting the Node-RED dashboard, it should be able to determine the camera type and optical characterization information without having to first restart the hardware controller.

Additional context
This issue is a follow-up on #210 (comment)

[ethanjli]

Because the ability to restart the Node-RED dashboard is buried in the Cockpit GUI (which is currently listed as being for advanced users), it is not yet considered an officially-supported functionality. This, this issue should not be encountered by most users. However, we will need to be careful about this issue if we instruct users to restart the Node-RED dashboard.

[ethanjli]

Copying in my analysis from #210 (comment):

Ultimately, a proper fix to this issue would be to allow the Node-RED dashboard to query the state of the hardware controller. But the MQTT API is not well-suited to this, though it is possible. Ideally we'd have a REST API for that kind of thing, but we may still need some intermediate solution with our MQTT API.

In the meantime, a reasonable workaround which doesn't touch the MQTT API could be to add a message somewhere visible in the Node-RED dashboard indicating that the hardware controller (or the PlanktoScope as a whole) needs to be restarted.

But it may be simpler to just add a request/response handler to the hardware controller allowing the MQTT API to query its camera type. Then the Node-RED dashboard can send a query message when it starts.

[ethanjli]

Note: this issue will also interact with #190, because (unless we modify the behavior of the MQTT API or the Python backend for reporting camera type) we won't be able to guarantee that the Node-RED dashboard is ready to receive MQTT messages by the time the Python backend sends its MQTT message reporting camera type.

We could have a fine-grained request/response protocol for each MQTT message type. Or we could just have a request from Node-RED at startup asking the Python backend to dump its entire state (including camera type) over MQTT.

[ethanjli]

Probably the simplest solution to this will just be to add an HTTP API with a GET route to check the camera model. We will anyways be adding an HTTP API POST route to update the system time, as a follow-up to #346.

[ethanjli]

Since @BiyouS is gradually working towards a complete rewrite of the Node-RED dashboard which might solve this problem, I will unassign myself from this issue.