feat: always infer function call when args empty or have comma

[gwhitney]

This is a proposed PR to resolve #3358. It interprets EXPR() and EXPR(..., ...) as function calls regardless of what EXPR is, rather than syntax errors unless EXPR is a symbol or accessor. To enable this syntax to cover unary function calls as well, it allows a trailing comma in argument lists, meaning exp(3,) is now a legal synonym for exp(3). With these conventions, you can write (f(x) = x+1)(6,) to get 7.

I would totally be open to other syntaxes that would allow one to call a function that was the result of an expression (rather than a direct symbol or accessor), but I wanted a PR available so that you could try this one in practice.

Note that re-allowing expressions that return functions to be called makes a small handful of the security tests no longer simply syntax errors. However, all such tests still show these exploits ineffective, due to other security measures (that were likely already in place when the prior syntax for calling a function returned from an expression, (EXPR)(3), was eliminated in favor of implicit multiplication).

Looking forward to your thoughts.

[gwhitney]

I recommend reading #3379 as well before reviewing this.