Compiler could warn for more unreachable int and float patterns

[fruno-bulax]

Currently in cases like these the second branch is not marked as unreachable:

case n {
  10 -> todo
  1_0 -> todo
}

case x {
  1.0 -> todo
  1.00 -> todo
}

This happens because the comparison is done on the raw string value rather than a parsed number. Implementers may want to peek at #5001 which tackled the same issue for floats in useless comparison warnings (1.0 == 1.00)

There exists a separate issue for bit array patterns: #5013

[lpil]

Oh yes! Good call.