optimize processing/parsing (especially for string input) (#62)

* move fast reading from tokenizer to processed input stream
* add more fast read bypass (tag name, attribute name/values)
* add custom logic for the string input case
* misc cleanup

Author: syjer

src/main/java/ch/digitalfondue/jfiveparse/Attributes.java

@@ -70,7 +70,7 @@ Set<String> keySet() {
 
     private void ensureMap() {
         if (attributes == null) {
-            attributes = new LinkedHashMap<>();
+            attributes = new LinkedHashMap<>(8);
         }
     }
 

src/main/java/ch/digitalfondue/jfiveparse/Characters.java

@@ -22,16 +22,21 @@ final class Characters {
 
     static final char NULL = 0x0000;
     static final char EXCLAMATION_MARK = 0x0021;
+    /** & */
     static final char AMPERSAND = 0x0026;
+    /** < */
     static final char LESSTHAN_SIGN = 0x003C;
+    /** > */
     static final char GREATERTHAN_SIGN = 0x003E;
     static final char SOLIDUS = 0x002F;
     static final char QUESTION_MARK = 0x003F;
 
     static final int EOF = -1;
 
     static final char TAB = 0x0009;
+    /** \r */
     static final char CR = 0x000D;
+    /** \n */
     static final char LF = 0x000A;
     static final char FF = 0x000C;
     static final char SPACE = 0x0020;

src/main/java/ch/digitalfondue/jfiveparse/EntitiesPrefix.java

@@ -19,7 +19,6 @@
 import java.io.IOException;
 import java.util.ArrayList;
 import java.util.List;
-import java.util.Map;
 import java.util.TreeMap;
 import java.util.zip.GZIPInputStream;