Planned or possible future developments for mrcfile.py
======================================================

* Lazy loading for high-performance filesystems - like mmap, but using much
  larger chunk sizes (mmap uses 4 KB chunks which cause inefficent access on
  high-performance parallel filesystems).

* Fixing of invalid files

* Handling of additional extended header types

* Data axis swapping (would need clarity about whether header fields such as
  cell size and dimensions apply to axes before or after swapping)
