Regression for type inference/ method overload in `katrix/dataprism`

[WojciechMazur]

Based on OpenCB failure in katrix/dataprism - build logs

Compiler version

Last good release: 3.6.0-RC1-bin-20240922-22ed2fb-NIGHTLY
First bad release: 3.6.0-RC1-bin-20240923-e5f7272-NIGHTLY
Bisect points to 9f90ad0

Minimized code

sealed abstract class Gen[+T] 
given [T]: Conversion[T, Gen[T]] = ???

trait Show[T]
given Show[Boolean] = ???
given [A: Show, B: Show, C: Show]: Show[(A, B, C)] = ???

object ForAll:
  def apply[A1: Show, B](f: A1 => B): Unit = ???
  def apply[A1: Show, A2: Show, B](f: (A1, A2) => B): Unit = ???
  def apply[A: Show, B](gen: Gen[A])(f: A => B): Unit = ???
  
@main def Test =
  ForAll: (b1: Boolean, b2: Boolean, b3: Boolean) =>
    ???

Output

[error] ./test.scala:13:8
[error] No given instance of type Show[(Boolean, Boolean, Boolean) => Nothing] was found for a context parameter of method apply in object ForAll
[error]     ???

Expectation

Should compile

Case 2

object ForAll:
  def apply[A1, B](f: A1 => B): Unit = ???
  def apply[A1, A2, B](f: (A1, A2) => B): Unit = ???
  
@main def Test =
  ForAll: (b1: Boolean, b2: Boolean, b3: Boolean) =>
    ???

[error] ./test.scala:21:3
[error] None of the overloaded alternatives of method apply in object ForAll with types
[error]  [A1, A2, B](f: (A1, A2) => B): Unit
[error]  [A1, B](f: A1 => B): Unit
[error] match arguments ((Boolean, Boolean, Boolean) => Nothing)
[error] 
[error] where:    A1 is a type variable
[error]           A2 is a type variable
[error]           B  is a type variable
[error]   ForAll: (b1: Boolean, b2: Boolean, b3: Boolean) =>
[error]   ^^^^^^