Use memory-mapped files in the data service

[rbs-jacob]

What is the use case for the feature?

OFRAK currently stores all data in-memory. This can be prohibitive for analyzing very large (>16GB) files, especially on memory-constrained systems.

How would you implement this feature?

Either modify the existing data service, or make an alternative implementation ascribing to the same interface. Instead of storing data as bytes objects, use memory-mapped files via Python's standard library mmap module.