add a way to specify the worker memory directly

[keewis]

The current way to control the worker memory is to define the desired worker memory, determine how many workers would fit into a single job, and specify that as the number of workers.

With this, instead, we specify the number of workers per job and automatically get the number of workers that fit into a single job.

A usage example would be:

cluster = dask_hpcconfig.create_cluster("name", **{"cluster.memory_limit": "4GB"})

we can also explicitly set the number of workers:

cluster = dask_hpcconfig.create_cluster(
    "name",
    **{"cluster.memory_limit": "4GB", "cluster.processes": 12},
)