Expose all synthetic semantic trees

[ingarabr]

The optimized lookup only exposed the last tree for a given position. A position can have multiple synthetic semantic trees. This occurs when we have a function with only an implicit argument.

Given the code:

object MyObject {
  trait Value[A] { def value: A }
  implicit val intValue: Value[Int] = new Value[Int] { def value: Int = 1 }
  class Foo[A]() {
    def baz[AA >: A](implicit s: Value[AA]): AA = s.value
  }

  val foo = new Foo[Int]()
  foo.baz
}

The tree for the foo.baz looks like this:

Term.Select(Term.Name("foo"), Term.Name("baz"))

When calling term.synthetic on the Term we're not able to extract the implicit arguments. If we add parentheses, foo.baz(), it will be wrapped in a Term.Apply, where the implicit is available. When querying the semantic tree from the given position we're ending up with two results where scalafix has optimized the lookup and expects at most one result.

This change exposes all the semantic trees that are available in the given position. It also depricates the API where we're expecting zero or one result.

[olafurpg]

Thank you for this contribution! LGTM 👍

[olafurpg]

This fix is now available in 0.9.7, I've kickstarted a release of the sbt plugin which may take an hour to finish

[ingarabr]

@olafurpg, thanks.