mfs: meta issue for future refactors

[aschmahmann]

From what I can tell given prior issues and PRs in go-mfs it seems like this package is due for a larger refactor rather than smaller changes.

Some issues for historical context for when someone is interested in refactoring (or writing an alternative) to mfs.

• Interface

  • Review the MFS interface go-mfs#3
  • move shouldn't need the root of the MFS as an argument go-mfs#20
  • Use FSNode's Type method instead of casting it to a File or Directory go-mfs#22
  • Make Lookup a member of Root and collapse it with DirLookup go-mfs#31
  • Evaluate the possibility of using an event handler based system, not a republisher based system go-mfs#33
  • [WIP] Clarify the motivation behind the flush option go-mfs#37
  • Clearly distinguish and document the two update mechanism in MFS: single-file upwards (childCloser) and full-update downwards (GetNode/sync) go-mfs#40
  • Review the MFS "cache" (it's not a cache) go-mfs#54

• Bug/Feature

  • Add support for symlinks in files go-mfs#4
  • Mv bug go-mfs#18
  • MFS root shouldn't need to put the node in the DAG go-mfs#34
  • crashing MFS with non-cached MerkleDAG go-mfs#83

If someone wants a particular issue transferred here just comment, however these issues are largely historical at this point and future refactors/rewrites may go in different directions which are also fine to discuss here 😄