docs: Write tuning guide as part of the user guide

[andygrove]

What is the problem the feature request solves?

We need to show users how to tune the performance of Comet. We do already have documentation for our configuration settings but we need a guide as well on how to use these.

Here are some thoughts on things that we should document:

• Columnar shuffle vs native shuffle
• Configuring number of parallel I/O threads when reading Parquet
• Configuring batch_size

Describe the potential solution

No response

Additional context

No response