Ref/vec

[sebffischer]

(Description is still WIP)

Overview of Changes:

• The Rep<T> now allows for non-atomic Ts. Specifically this now allows List to be simply represented as an Rep<T>. Furthermore this will also support named vectors because the (optional) naming mechanism was integrated into the Rep struct.
• Some refactors were done to push Rep<T> towards a more stable API. E.g. it is now possible to iterate over immutable references of the vector representation, representing the subsets. This can be done via rep.pairs_ref().iter() (.pairs_ref_iter() would not work due due lifetime issues and the restrictions of the netxt(&mut self) signature in the Iterator trait; I think this might be achievable with unsafe but I want to refrain from this for the time being)
• Added the is_null(x) and all(...), is_na(x) primitives that allow for better testing through via the R interpreter (r_expect!, r!). The documentation of all includes a section on 'Deviations from R'.
• Setting a value in a list to null now does not remove this element anymore but actually sets it to null.

Bug Fixes:

• [[.<- assignment has been fixed: Bug when accessing nested list #179 (comment)
• The length() function now works for lists.
• Concatenating lists via c() now works as expected: (c() primitive behaves differently than in R when coercing to a list #177). E.g. c(list(1), list(2)) evaluates to list(1, 2).
• names() now respects subsets: names() does not respect subset #181

---

Open Issues / Questions:

• $.<- assignment was broken for environments by this PR
• use internal_err!() instead of unimplemented!() where possible
• unify $ and [[ internally. I think this should also fix: Bug when accessing nested list #179 (comment)
• ensure and test that named vectors can be created
• fix: c() primitive behaves differently than in R when coercing to a list #177
• decide whether we want to allow subsetting with NA. --> not in this PR.
• fix list(1)[[na]] and list(1)[na] will do this in a new PR
• add tests for the different iterator methods of reptype
• update CHANGELOG.md

---

Next (bigger) steps

I would plan to continue focusing on the vector representation to polish the internals and further improve the behavior.
This includes:

• Enforcing proper recycling rules.
• I think it would also be nice to continue the work with the scalars WIP: Adding scalars and scalar indexing #135 to push the vector-internals further into a more stable API. But this is of course your call. I could also offer to help there if you are currently busy with other things :)

[dgkf]

Let me know when you're ready for review