Reduce overhead of monitor pipeline

[watson]

This PR optimizes two things:

Use pipeline over pumpify

This is mainly to fix an "issue" with Node.js 14 where pumpify requires a few extra ticks every time someone writes to the pipeline. This could result in the process crashing prior to the logger writing to its destination. For example, in the following example, the data event-listener would never be called before the process crashes because of the thrown error on Node.js 14:

const Pumpify = require('pumpify')
const { PassThrough } = require('stream')
const a = new PassThrough()
const b = new PassThrough()
const pipe = Pumpify([a, b])
b.on('data', (chunk) => console.log('Got data:', chunk.toString()))
pipe.write('hello')
throw new Error('boom!')

However, the following would call the data event-listener on Node.js 14:

const { PassThrough, pipeline } = require('stream')
const a = new PassThrough()
const b = new PassThrough()
pipeline(a, b, () => {})
b.on('data', (chunk) => console.log('Got data:', chunk.toString()))
a.write('hello')
throw new Error('boom!')

Don't use pipeline if not required

In the case where there's only one stream, there's no need to use a pipeline. Instead we can just write directly to the stream. This also reduces the overhead of the logger slightly.

[spalger]

LGTM

[spalger]

Nice, didn't know about the finished() helper!

[mshustov]

OMG, finally it's provided out-of-the-box

[spalger]

Oops, missed the test failure

[mshustov]

Considering that @hapi/good is EOL on 31.12.2020, I'm okay with this divergence from the upstream. I created elastic/kibana#82240 to remove the dependency on @hapi/good package.

[spalger]

LGTM

[watson]

@spalger my latest commit only fixed part of the CI issue. There's still some failures. However, it looks like master is also failing currently. That being said, it also looks like the failures in this PR is slightly different from the ones in master, so I'll take a look tomorrow to see what's going on.

[watson]

Sorry for messing up the PR with a lot of unrelated commits. I just rebased this PR on top of #6 to get a more stable CI. Once #6 is merged into master, I'll rebase this PR with master to clean it up.

[spalger]

LGTM, thanks again for fixing CI!