feat(deduplication): ignore periods when comparing names

This is a small extension to our deduplication logic so that periods in
names will be ignored when comparing for dedupication.

For example, a query for `3929 St Marks Avenue, Niagara Falls, ON,
Canada` returns two duplicate addresses from OpenAddresses. One is
sourced from a countrywide dataset, and another a regional dataset. One
has a period after the abbreviation for Saint, one doesn't.

We should probably evaluate ignoring most or all punctuation, but this
fixes a somewhat common case for now.

Author: orangejulius

helper/diffPlaces.js

@@ -386,7 +386,7 @@ function layerDependentNormalization(names, layer) {
  * lowercase characters and remove diacritics and some punctuation
  */
 function normalizeString(str){
-  return removeAccents(unicode.normalize(str)).toLowerCase().split(/[ ,-]+/).join(' ');
+  return removeAccents(unicode.normalize(str)).toLowerCase().split(/[ ,-.]+/).join(' ');
 }
 
 module.exports.isDifferent = isDifferent;

test/unit/helper/diffPlaces.js

@@ -737,6 +737,7 @@ module.exports.tests.normalizeString = function (test, common) {
     t.equal(normalizeString('foo, bar'), 'foo bar');
     t.equal(normalizeString('foo-bar'), 'foo bar');
     t.equal(normalizeString('foo , - , - bar'), 'foo bar');
+    t.equal(normalizeString('St. Marks'), 'st marks');
     t.end();
   });