Add recipe on converting a large CSV into a smaller one

Add a recipe which shows how to do `open_dataset()`...`write_dataset()`, and some extra text which explains why this is better than doing head/tail/awk from one giant csv to many smaller ones (i.e. things will go wrong with headers present or missing.)

Start by locating the repository's existing recipe documentation and the entry points named in the issue: open_dataset() and write_dataset(). Add a recipe that demonstrates converting one large CSV into smaller ones, and explain why this avoids header problems that can occur with head, tail, or awk.

Written by the indexing model from the issue text.