[#56] Use || instead of && in the Scala jw UDF

This doesn't affect the return values of the function, but it does mean
that we only call JaroWinklerSimilarity.apply() when both arguments are strings
with length > 0. This should help prevent similar edge cases from showing up
in the future.

scala_jar/src/main/scala/com/isrdi/udfs/SerJaroWinklerSimilarity.scala

@@ -7,8 +7,10 @@ package com.isrdi.udfs
 import org.apache.commons.text.similarity._
 
 class SerJaroWinklerSimilarity extends JaroWinklerSimilarity with Serializable {
-  // JaroWinklerSimilarity.apply("", "") returns 1.0, but for our use case it makes
-  // more sense for two empty strings to have similarity 0.0.
+  // If either input string is empty, immediately return 0. Otherwise, return
+  // the value of calling JaroWinklerSimilarity's apply function.
+  // JaroWinklerSimilarity.apply("", "") returns 1, but for our use case it makes
+  // more sense for two empty strings to have similarity 0.
   //
   // By my understanding, the comparison of two empty strings is essentially
   // undefined. To me it makes sense to directly apply the definition of
@@ -17,7 +19,7 @@ class SerJaroWinklerSimilarity extends JaroWinklerSimilarity with Serializable {
   // empty strings 1 - 0 = 1. So I understand the decision to make the
   // similarity 1 and the distance 0 as well.
   override def apply(left: CharSequence, right: CharSequence): java.lang.Double = {
-    if (left == "" && right == "") {
+    if (left == "" || right == "") {
       0.0
     } else {
       super.apply(left, right)