refactor: use Tuple instead of Struct in saw-remote-api backend

[pnwamk]

Use the more generic "tuple" instead of "struct" notion of value in the server backend (while keeping "struct" in the llvm-interface). See #1031

[RyanGlScott]

Does this imply that we should rename the data structures corresponding to the struct type (in addition to the struct setup value)?

[pnwamk]

Does this imply that we should rename the data structures corresponding to the struct type (in addition to the struct setup value)?

I'm re-reading @atomb's original comments on the matter. The impression I got was the things which are not inherently LLVM-related should prefer the more generic "tuple", while things that are LLVM specific should be free to use the terminology/type names/etc from LLVM. To that end, I think it's right to rename the SetupVal to "tuple" because it is a crucible/core concept, whereas the types are LLVM specific, so they should keep "struct" as the name.

[RyanGlScott]

I know that this isn't the officially sanctioned source of documentation anymore, but I'd find it helpful if old-SAW.rst were updated to reflect this change to the SAW remote API, especially in light of #1121.

[atomb]

This looks good to me. And, yes, the idea was that language-agnostic stuff would use tuple/record names whereas language-specific stuff would use language-specific names (like struct). At the moment, that distinction is somewhat academic, but once the Java bindings are in place I think it'll be more meaningful.