Add rvalue get for tuples

[MrBurmark]

Here is my attempt to add support for rvalue forms of get with tuple.
This fixes #116.

[trws]

I quite like this, but we need some tests here to ensure that the rvalue getters work as necessary. Probably not a ton to do there, but at least some basic "make_tuple(0).get()" or similar for a few different situations so we have a smoke test for this would make me a lot more comfortable it wont accidentally be broken down the line.

[MrBurmark]

Of course, I'll add tests.

[MrBurmark]

One thing to note about this is that it requires changes to be made in RAJA, so it will cause a break in compatibility so this change can only be made when we update everything. Do we want to add this for the immanent release or wait until a later release?

[rhornung67]

I think we should get these changes in for the release.

[trws]

Does it break any external APIs in raja? It's an ABI break because of the new overloads but if the API level at least at raja is ok I'm a lot less worried about the fallout.

[MrBurmark]

It breaks stuff related to RAJA::zip_tuple. We have to get rid of the RAJA::get implementation here https://github.com/LLNL/RAJA/blob/develop/include/RAJA/util/camp_aliases.hpp#L69 which probably should never have existed in the first place. We also have to refactor the RAJA::get for zip_tuple https://github.com/LLNL/RAJA/blob/develop/include/RAJA/util/zip_tuple.hpp#L47.
I don't think anyone is using this stuff directly.

[trws]

That seems pretty reasonable.

[trws]

@MrBurmark, do you consider this good to go? Tests look pretty good, I would normally include some actual executable tests in with the static asserts but you have really good coverage there so we might be alright in this case.

[MrBurmark]

I'll add some executable tests if its what you would do.

[MrBurmark]

@trws I'm ok with this once the tests pass.

[trws]

Looks great!