[Elasticsearch-UI][Watcher] Threshold graph in watcher scale is not accurate

[BobBlank12]

Kibana version: 6.6.0

Elasticsearch version: 6.6.0

Server OS version: Mac 10.14.3

Browser version: Chrome 73.0.3683.86 (Official Build) (64-bit)

Browser OS version: Mac 10.14.3

Original install method (e.g. download page, yum, from source, etc.): tar.gz

Describe the bug: When creating a threshold alert in watcher, the scale on the graph in the GUI does not match the actual events from the same query against Elasticsearch. Both examples below are 5 second buckets.

Steps to reproduce:

1. Created a consistent data feed using logstash generator input
2. Used Kibana Discover to get accurate count of events in 5 second buckets
3. Created Threshold watch in Watcher through GUI utilizing graph to determine a good threshold to start with. 5 second buckets were displayed.

Expected behavior: The graph in the Threshold watch should have a similar event count as seen in Discover. You can see from the images below, each using logstash* as the index pattern (this was the ONLY logstash feed going into the cluster) that the Discover app shows on average 300,000 documents per 5 second bucket while the graph in the Threshold Watcher UI shows around 70,000 documents per 5 second bucket.

Screenshots (if relevant):

Errors in browser console (if relevant): N/A

Provide logs and/or server output (if relevant): See above screen shots.

Any additional context: The graph in watcher doesn't make it easy to choose an accurate starting point to create a threshold watch as is.

[elasticmachine]

Pinging @elastic/es-ui