Add in_process_executor use-case to Kubernetes deploy documentation.

[czroth]

Summary & Motivation

We have a partitioned job running on our Kubernetes cluster which benefited greatly by switching from the default executor to the in_process_executor.

Other Kubernetes deployed users may benefit from the same setup and this in not covered in the kubernetes deploy docs.

In our case, we went from this being typical:

to this:

Resulting in our jobs running in a third of the time, and using fewer resources.

Steps like the following were completely avoided by switching to the in_process_executor:

How I Tested These Changes

By deploying the user code to our production server with only the single change from the default executor to the in_process_executor.

[czroth]

@erinkcochran87, thanks for picking this up to review. If you need any additional details from me, please ask.

[czroth]

h/t to @sebaap for helpful discussion related to this PR on Dagster's slack.

[erinkcochran87]

Thanks for adding this! I'm looping in @gibsondan and @mlarose to also give this a once-over.

Could you also please run make mdx-format in the /docs directory and commit any changes? The Markdown files in our docs have be formatted a specific way to pass our tests - this command will do the formatting for you.

[czroth]

@mlarose , I believe I've addressed your comments. Let me know if I missed anything.
@erinkcochran87 , let me know if it reads better with less repetition on my part.

In the last commit, I combine two paragraphs that start with the same sentence and come on either side of the heading Kubernetes configuration on every step in a run. I think this reads better than having them separate, but if you disagree, let me know and I'll revert the change.

[czroth]

And I re-ran yarn mdx-format before the last commit. But I've never used mdx or yarn before, so I'm uncertain as to whether I've got everything correct here, including the PyObject module links. Someone will need to double check that.