[Monitoring] Logstash pipeline listing page show inaccurate rate data

[chrisronline]

If you look at the pipelines overview page and note the events emitted rate, then go to the pipelines listing page and add up all the events emitted rate, the numbers are not the same (and not even close).

There are two things causing this:

• This uniq is actually filtering our valid hash and ephemeral_id values that are necessary in the subsequent queries to collect all the data
• Subtracting min from max leaves potential gaps of data that isn't included in the calculation in smaller bucket sizes - when looking at min/max in individual buckets, the max of the first usable bucket doesn't always equal the min of the next usable bucket and this becomes more prevalent when the number of documents per bucket is small.

We need to make two fixes:

1. Fix the bug with the uniq above so it properly groups sets of hash and ephemeral_id values with each unique id

2. Change how we collect the pipeline metric data for the listing page. Instead of a single query, we should issue multiple queries in parallel that only look at a single pipeline id (since we now have the list up-front). We want to rely on a ES derivative to do the calculations (to avoid the issues mentioned above) but in order to do that with the way logstash monitoring documents are structured, we need to do separate queries for individual pipeline ids.

cc @inqueue

[elasticmachine]

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