Shrink integers toward zero instead of typemin

[jariji]

It seems that integers currently shrink toward typemin(Int):

julia> @check ne_self(x=Data.Integers{Int}()) = x ≠ x
Found counterexample
  Context: ne_self

  Arguments:
      x::Int64 = -9223372036854775808

I have trouble reading these many-digit numbers, so it's hard to tell if an assertion like -9223372036854775808 < -9223372036854775808 is correct or not. I think I would have an easier time making sense of the errors if they shrank toward zero instead so that the assertions were more like 0 < 1.

[Seelengrab]

Yes, this is definitely a nice to have. The code currently responsible for producing integers is the same for signed & unsigned types, located here. This would need to be specialized for signed integers, such that the encoding for negative numbers is considered strictly larger than the encoding for positive numbers.