[Stack Monitoring] Shard size alert never fires

[simianhacker]

Kibana version:
master (probably more)
Elasticsearch version:
master (probably more)
Original install method (e.g. download page, yum, from source, etc.):
source
Describe the bug:
When you set the shard size threshold to zero the alert never fires.
Steps to reproduce:

1. Start a cluster with internal monitoring
2. Create default alerts
3. Set the shard size threshold to zero
4. Alert never fires

Expected behavior:
The alert should fire
Any additional context:
It looks like the source_node information is missing from the index_stats documents. This is causing the alert be skipped: https://github.com/elastic/kibana/blob/master/x-pack/plugins/monitoring/server/lib/alerts/fetch_index_shard_size.ts#L136-L143

[elasticmachine]

Pinging @elastic/stack-monitoring (Team:Monitoring)

[neptunian]

I think from what @simianhacker and I discussed this was working with internal monitoring but NOT metricbeat monitoring due to fields possibly removed from metricbeat or never existing