Move output endpoint from Che master container to a separate one

[garagatyi]

Description

When we were designing Che 6 we added an endpoint for pushing different kinds of output from a workspace.
It was supposed that we can move this out of the Che master for the scalability reasons.
Now we see the significant load from logs which leads to OOMs of Che master.
So, we finally need to do this move.
We will need to add it for logs ONLY, not for the cases when we sent statuses.
We will need to add separation of statuses/logs to the plugin broker.
Broker: link
Installer: link, link, link, link

Reproduction Steps

OS and version:

Diagnostics:

[garagatyi]

@skabashnyuk @sleshchenko FYI

[garagatyi]

It looks like clients do not use links they were supposed to use to get logs from Che master. Instead of that, they construct websocket link themselves and use it. The link they construct happens to be correct for now just because of the behavior of JSON_RPC that a client can connect to any link and use any JSON_RPC method.
The correct way of getting logs is to get the link with name from workspace retrieved from Che master.
Here is how workspace-loader constructs a link instead. I believe that we have the same issue for factory accepting page (with the crane) and maybe some others which I can't recall at the moment.
I believe that we will need to separate logs brokering to a separate service we need to start work on fixing clients ASAP.
@ashumilova @evidolob @l0rd

[garagatyi]

I experimented with it a bit. Here are the changes in Che plugin broker to use a separate endpoint for the output https://github.com/eclipse/che-plugin-broker/compare/logsEndpoint?expand=1
Here are basic changes in Che https://github.com/eclipse/che/compare/master...garagatyi:outputPropagator?expand=1. But we stream some logs directly from Che master here and here and here
We would need to either send those logs to a remote service from Che master or stop streaming them at all.
Here is a sample of such a service I prepared. But it doesn't start correctly for some reason and I haven't figured out why yet.

[che-bot]

Issues go stale after 180 days of inactivity. lifecycle/stale issues rot after an additional 7 days of inactivity and eventually close.

Mark the issue as fresh with /remove-lifecycle stale in a new comment.

If this issue is safe to close now please do so.

Moderators: Add lifecycle/frozen label to avoid stale mode.