Make is_syntactic_literal() more strict

[jonthegeek]

Add an attributes() check to is_syntactic_literal() to ensure that syntactic-literal-like objects such as factors and arrays are not treated as syntactic literals.

Fixes #1817.

[jonthegeek]

#1799 is really close to this. Would you like me to fix that in this same PR, or keep it separate? It will make it make sense to split integer and double off from character (possibly with a shared helper), so the structure of what I did here might change a bit. In any case it doesn't feel like it makes sense to work on the separate PR until this one is reviewed.

Edit: Technically character can remain as-is with just length(x) == 1, but it's probably technically better for it to also have the is.null(attributes(x)) check, and it feels like a shared helper is good for those in case anything else comes up along the same lines.

[lionel-]

A second PR for the other fix would be great, thanks!