Allow relative precision in equality comparison

[nightscape]

Hi @MrPowers,

I'd like to switch over from spark-testing-base to spark-fast-tests in spark-excel.
I'm using property-based testing to cover a wide range of edge cases when writing and reading Excel files.
The problem with this is that I don't (want to) have control over the range of values for e.g. a double column.
When using assertApproximateDataFrameEquality I can only specify an absolute precision which would be too big for some random numbers and too small for others.
Would you be open to a PR that adds this as additional capability?
I might be able to do this in a source- but not binary-compatible way sth. like this:

trait NumericComparison {
  def compare[T : Numeric](t1: T, t2: T): Boolean
}

case class AbsoluteNumericComparison(precision: Double) extends NumericComparison {
  def compare[T : Numeric](t1: T, t2: T): Boolean = ???
}

case class RelativeNumericComparison(precision: Double) extends NumericComparison {
  def compare[T : Numeric](t1: T, t2: T): Boolean = ???
}

object NumericComparison {
  implicit def doubleToAbsoluteNumericComparison(precision: Double) = AbsoluteNumericComparison(precision)
}

def assertApproximateDataFrameEquality(df1: DataFrame, df2: DataFrame, precision: NumericComparison) = ???

This would give room for further extensions to comparisons.

A simpler alternative might be

def assertApproximateDataFrameEquality(df1: DataFrame, df2: DataFrame, precision: Double, precisionIsRelative: Boolean = false) = ???

Do you think this could work, or do you see another way to add relative precision?