Feature: use PyFilesystem zipfile instead of zipfile
Reading and writing workbooks uses a lot of CPU and, with a lot of globals bound to the workbook object, it's not really possible to do any parallelisation.
But the PyFilesystem Zipfile implementation could simplify a lot of read/write code and might also make it possible to at least use async in some options, though a separate process would be even better: this could make writing very large workbooks noticeably faster.