Improved synonyms memory footprint.

The idea is to avoid unnecessary String objects duplication for the same
words. Particularly efficient with the large moby thesaurus.

Memory footprint measurements with VisualVM :
 - openthesaurus_de_yacy :
 	- initial : 19 443 796 bytes
 	- after refactoring : 18 012 606 bytes

 - mobythesaurus_en_yacy :
 	- initial : 343 453 904 bytes
 	- after refactoring : 173 843 780 bytes

 - thesaurus_ru_yacy :
 	- initial : 3 800 706 bytes
 	- after refactoring : 3 466 612 bytes

 - de + en + ru : 
 	- initial : 366 603 450 bytes
 	- after refactoring : 195 015 914 bytes

source/net/yacy/cora/language/synonyms/SynonymLibrary.java

@@ -49,6 +49,9 @@ public class SynonymLibrary {
         lib.clear();
         if (!path.exists() || !path.isDirectory()) return;
         final String[] files = path.list();
+        /* Global map of all known distinct words : thus enable reuse of the same word String instance
+         * appearing multiple times in different synonyms sets */
+        final Map<String, String> distinctWords = new HashMap<>();
         for (final String f: files) {
             File ff = new File(path, f);
             String line;
@@ -62,12 +65,20 @@ public class SynonymLibrary {
                     String[] words = CommonPattern.COMMA.split(line);
                     Set<String> synonyms = new HashSet<String>();
                     Set<String> keys = new HashSet<String>();
-                    for (String s: words) {
-                        s = s.trim();
-                        if (s.length() < 2) continue;
-                        String t = s.toLowerCase();
-                        synonyms.add(t);
-                        keys.add(t.substring(0, 2));
+                    for (String word: words) {
+                        word = word.trim();
+                        if (word.length() < 2) continue;
+                        String lowCaseWord = word.toLowerCase();
+                        String kownWord = distinctWords.get(lowCaseWord);
+                        if(kownWord != null) {
+                        	/* This word is already known : let's use the existing String instance from the synonyms map to gain memory space */
+                        	lowCaseWord = kownWord;
+                        } else {
+                        	/* First encounter of this word : let's add it to the global map of known words */
+                        	distinctWords.put(lowCaseWord, lowCaseWord);
+                        }
+                        synonyms.add(lowCaseWord);
+                        keys.add(lowCaseWord.substring(0, 2));
                     }
                     for (String key: keys) {
                         List<Set<String>> symsetlist = lib.get(key);