In #8537, among other things, InMemoryLayer learned to batch its IOs into aligned+sized chunks.

In #8763 , vectored reads learned to do a similar thing.

This task is to:

• explore whether there are opportunities for sharing code and tests
• do the work if it makes sense.

Don't jump at implementing, do exploration first and post a report as a comment to this thread.

To be done after both of these PRs have merged, shipped, and proved themselves in practice.