Fix valid-let-bindings?

[jpellegrini]

We had this:

(let 1 2)
=> #[primitive %execute]

because valid-let-bindings? would return #f and not trigger an error when bindings was not a list.

This patch fixes that error, and passes all tests.

[jpellegrini]

By the way, I'm also working on SRFI 71 (Extended LET-syntax for multiple values), and that's how I found this problem.
71 is not a popular SRFI, and I know we already have let-values, but... It's formally a standard, and actually not that hard to implement. :)

[jpellegrini]

By the way, I didn't include tests because the error raised cannot be caught... It would break the test procedure.
Is there a way to catch any possible error, including those during compilation?

[jpellegrini]

Perhaps it would also make sense to change the procedure name to check-let-bindings, since after this patch it would not be a predicate anymore?

[egallesio]

Thanks @jpellegrini, for this PR.

Is there a way to catch any possible error, including those during compilation?

Yes, use test/compile-error.

Perhaps it would also make sense to change the procedure name to check-let-bindings, since after this patch it would not be a predicate anymore?

You are right. Done.