Per query resource allocation (i.e. evenly split memory and CPU between queries)

At the moment it is possible to specify per-executor memory limits. Add support for another mode for resource allocation - per-query. 

This will be the mode whereas both CPU and memory is going to be evenly divided for available concurrency. For CPU this can be either achieved via through `target_partitions` settings or limiting thread pool for `DedicatedExecutor`.

