Implement IOList BIFs

[mlwilkerson]

Resolves #153.

• Implement and test iolist_to_binary/1
• Implement and test iolist_size/1
• Implement and test iolist_to_iovec/1
• Add testing for SubBinary
• Figure out whether to add testing for MatchContext
  This doesn't seem to be possible yet.

[mlwilkerson]

I'm starting with pretty naive implementations here, to get my feet wet. This is my first Rust programming, and first exposure to both the Lumen dev environment and to iolist, so I've gotta find a handhold somewhere.

It's taken a bit of digging to find good information on the use cases for iolist vs the similar list_to_binary/1. (My initial implementations delegate to list_to_binary/1.)

I'm beginning to understand the use case of optimizing memory usage when streaming data such as when writing to a file or socket. The iolist functions can allow for more efficient list operations and memory usage. Handling improper lists is part of this use case. Helpful resources on this: here, here, here, here.

[mlwilkerson]

I'm not sure whether adding test coverage for the binary sub-types is important. Interested to hear what a reviewer thinks about that.

[mlwilkerson]

@KronicDeth do you have any feedback regarding whether it's important to test with various binaries that would result in different internal typed terms? (ProcBin, HeapBinary, SubBinary, MatchContext)

Seems like I saw something possibly related on #200 .

[KronicDeth]

You definitely need to test subbinaries to show that you handle non-binary ones correctly.

[mlwilkerson]

I rebased onto develop and fixed up some stuff that seems to have changed.

I was not successful at constructing a MatchContext that I could pass in as a term from the test module.

I tried adding a test for ProcBin, which seemed to expose the fact that the list_to_binary/1 (to which iolist_to_binary/1 currently delegates). Opened #368 to track that and proposed #369 to resolve it. So the resolution of this PR is now blocking and that resolution of those.