max_threads_for_leaf setting

### Company or project name

_No response_

### Use case

Clusters with huge amount of shards.
Due model of how clustered queries are executed in ClickHouse:
Initial node may handle disproportionally big chunk of query execution pipeline.
It force users to set higher max_threads value to handle this load, but it's not useful for shards as processing parallelism depends first (and scale better) with amount of shards. So, it make sense to set max_threads to lower value for them (it will be multiplied by number of shards anyway).

### Describe the solution you'd like

Set different concurrency level for initial node and remote shards
(max_threads vs max_threads_for_leaf)

### Describe alternatives you've considered

_No response_

### Additional context

_No response_