Set default concurrency limit based on urgency and capacity

Originated from gitlab-org/gitlab!190565 (comment 2556618734).

We need to set a default concurrency limit for all workers as they will be subject to throttling.

When a worker class doesn't have a limit statically defined, it will follow a percentage of the shard's capacity. For example, we'd have a worker on catchall that would only be allowed to use 20% of total threads (2160 concurrent jobs currently). In formula:

(threads * maxReplicas) * maxPercentage

The maxReplicas can be exposed as an environment variable initially for .com. Once we want to support self-managed, it could be tied to routing rules.

Individual workers could adjust this percentage, and we could allow configuration to override the count altogether (#208 (closed)).

Tasks

1. Inject maxReplicas from sidekiq charts
2. Set the default concurrency limit based on urgency & capacity
   • Add a new percentage attribute to adjust the maxPercentage of individual worker