Fix the substrate networking Grafana dashboard.

[wi1dcard]

As mentioned in PR #7038 :

Most importantly, the list of nodes was fetched from what the cpu_percentage metric reports, and would therefore show as empty.

Since the sysinfo has been removed from the Prometheus metrics, the metric we used to extract instances ${metric_namespace}_cpu_usage_percentage no longer exists. However, #7038 only updated the substrate-service-tasks dashboard, this PR fixes the substrate-networking dashboard.

This PR also:

1. Changes the time range from 24h to 12h and disables auto-refresh, in order to be consistent with the substrate-service-task dashboard.
2. Updated the dashboard JSON to Grafana 7.0 (I'm also willing to submit another PR to update the substrate-service-task dashboard).

[cla-bot-2020]

@wi1dcard it looks like you have not signed our contributor license aggreement yet. Please visit this link to sign our agreement. This pull request cannot be merged until the agrement is signed.

[cla-bot-2020]

@wi1dcard, Your signature has been received.

[tomaka]

Hi!

Sorry for that, I was waiting for 0.8.24 to be widely deployed before updating this dashboard because some metrics have been modified (#6634) in that version. Releasing 0.8.24 took way more time than expected.

Do you mind if I open a PR myself that fixes the dashboard and also updates the metrics for 0.8.24?

[wi1dcard]

Hi!

Sorry for that, I was waiting for 0.8.24 to be widely deployed before updating this dashboard because some metrics have been modified (#6634) in that version. Releasing 0.8.24 took way more time than expected.

Do you mind if I open a PR myself that fixes the dashboard and also updates the metrics for 0.8.24?

Oh, thank you for reminding me, I did't even notice that there was another change of the metrics. It would be great if you can open another PR to fix the dashboard and update the metrics as well! Feel free to close this one.