Fix NGramFingerprintKeyer to ignore accents - fixes #1161 (#2899)

Fixes #1161
This change parallels what was done in #1257 1da3c00 to fix
the FingerprintKeyer and moves the diacritic removal before
the deduping. Includes a test.

main/src/com/google/refine/clustering/binning/NGramFingerprintKeyer.java

@@ -49,13 +49,14 @@ public class NGramFingerprintKeyer extends FingerprintKeyer {
         }
         s = s.toLowerCase(); // then lowercase it
         s = alphanum.matcher(s).replaceAll(""); // then remove all punctuation and control chars
+        s = asciify(s); // find ASCII equivalent to characters
         TreeSet<String> set = ngram_split(s,ngram_size);
         StringBuffer b = new StringBuffer();
         Iterator<String> i = set.iterator();
         while (i.hasNext()) { // join ordered fragments back together
             b.append(i.next());
         }
-        return asciify(b.toString()); // find ASCII equivalent to characters 
+        return b.toString();
     }
 
     protected TreeSet<String> ngram_split(String s, int size) {

main/tests/server/src/com/google/refine/clustering/binning/KeyerTests.java

@@ -62,9 +62,10 @@ public class KeyerTests extends RefineTest {
     
     private static final String[][] testNGramStrings = {
         {"abcdefg","abbccddeeffg"},
+        {" a,b.c d\te!f?g ","abbccddeeffg"},
+        {"écÉCec","ceec"},
         {"",""}, //TODO: add more test cases
         {"",""},
-        {"",""},
     };
     
     @Override