Determine number of workers and blocks to set for parsl config in GKE

[julietcohen]

Within the Kubernetes workflow, regardless if deploying on the ADC cluster or on GKE, the parsl config contains parameters for:

• max_blocks = 4,
• min_blocks = 1,
• init_blocks = 1,
• max_workers = 4,
• cores_per_worker = 1,

These parameters determine how the work is distributed across workers. Before re-building and pushing the image, the user should adjust these as needed depending on the size of the input dataset. Parsl's documentation for blocks is here.

@shishichen noted in the GKE documentation: "During the script execution, parsl will bring up additional worker pods as needed."

I confirmed this during my first small run of the GKE workflow. I set the parameters to the values above, and initially there were 4 connected parsl workers, then that jumped to 8. This is not an explicit problem, but ideally we would control this number closely with the config. We should add more documentation about when to use blocks versus workers.