Improve escape hatch for filter in generator

[som-snytt]

Allow for (x @ (_: T) <- ts) workaround using
underscore as shown. Because the rewrite happens
after patterns are transformed, it was taken as
a straight typed pattern.

[som-snytt]

The failing test is roundtripping in quasiquotes, which doesn't understand the partial function with Tuple1 decomposition.

Currently:

$ scalam
Welcome to Scala 2.13.0-M5 (OpenJDK 64-Bit Server VM, Java 1.8.0_191).
Type in expressions for evaluation. Or try :help.

scala> import reflect.runtime._,universe._
import reflect.runtime._
import universe._

scala> q"""for (x @ (y: String) <- List("")) yield x"""
res0: reflect.runtime.universe.Tree =
List("").withFilter(((check$ifrefutable$1) => check$ifrefutable$1: @scala.unchecked match {
  case (x @ (y @ (_: String))) => true
  case _ => false
})).withFilter(((check$ifrefutable$2) => check$ifrefutable$2: @scala.unchecked match {
  case (x @ (y @ (_: String))) => true
  case _ => false
})).map(((x$1) => x$1: @scala.unchecked match {
  case (x @ (y @ (_: String))) => x
}))

scala> for (x @ (y: String) <- List("")) yield x //print

scala.collection.immutable.List.apply[String]("").withFilter(((check$ifrefutable$1: String) => (check$ifrefutable$1: @scala.unchecked) match {
  case (x @ (y @ ((_): scala.Predef.String))) => true
  case _ => false
})).map[String](((x$1: String) => (x$1: @scala.unchecked) match {
  case (x @ (y @ ((_): scala.Predef.String))) => x
})) // : List[String]

[som-snytt]

Reviving to add pattern improvements under -Xsource:3, to support case in for, and also @unchecked ascription. Currently, lint of refutable pattern in vardef requires @unchecked on the rhs.

Jenkins, would you let me know which test was failing? My branch had a bunch of debug added for quasiquotes, but no test. ("Scalacheck tests.")

[som-snytt]

Would be nice to support implicit val _: Int = 42 (i.e., fresh name) under -Xsource:3.

And warn List(1,2,3).map { i: Int => i + 1 } (issue 12218)

[som-snytt]

Closing for indifference.

[SethTisue]

whose indifference? I'm not indifferent! (I think @dwijnand isn't either...?)

to be clear, I'm not claiming my non-indifference was externally observable

aligning with Scala 3 here, insofar as it proves to be possible, seems appealing

[dwijnand]

I think we should implement Scala 3's escape hatch rather than our own. So I'm not indifferent (I'm different) on the what but I'm not in favour of the how here. Unless I'm missing something?