Add Automata.makeCharSet/makeCharClass to optimize regexp (#14193)

Add Automata.makeCharSet(int[])/makeCharClass(int[],int[]) to optimize regexp.

* Add new "character class" node, which was previously composed by union
  of many nodes.
* Remove "predefined class" node, which previously built an internal
  separate regex on the fly, it is just another character class.
* RegExp no longer uses union() internally, except for union (|) operator.
* format codepoints in the internal parse tree output with U+%04X
* Fix concatenate to remove the dead states it creates, just like
intersection/union/etc do
* fix dead-states-test to explicitly create dead states, rather than
relying on some function to create a mess. it doesn't anymore.

lucene/core/src/java/org/apache/lucene/util/automaton/Automata.java

@@ -32,6 +32,7 @@
 import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Collection;
+import java.util.Objects;
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.BytesRefIterator;
 import org.apache.lucene.util.StringHelper;
@@ -140,6 +141,32 @@ public static Automaton makeCharRange(int min, int max) {
     return a;
   }
 
+  /** Returns a new minimal automaton that accepts any of the provided codepoints */
+  public static Automaton makeCharSet(int[] codepoints) {
+    return makeCharClass(codepoints, codepoints);
+  }
+
+  /** Returns a new minimal automaton that accepts any of the codepoint ranges */
+  public static Automaton makeCharClass(int[] starts, int[] ends) {
+    Objects.requireNonNull(starts);
+    Objects.requireNonNull(ends);
+    if (starts.length != ends.length) {
+      throw new IllegalArgumentException("starts must match ends");
+    }
+    if (starts.length == 0) {
+      return makeEmpty();
+    }
+    Automaton a = new Automaton();
+    int s1 = a.createState();
+    int s2 = a.createState();
+    a.setAccept(s2, true);
+    for (int i = 0; i < starts.length; i++) {
+      a.addTransition(s1, s2, starts[i], ends[i]);
+    }
+    a.finishState();
+    return a;
+  }
+
   /**
    * Constructs sub-automaton corresponding to decimal numbers of length x.substring(n).length().
    */

lucene/core/src/java/org/apache/lucene/util/automaton/Operations.java

@@ -148,8 +148,7 @@ public static Automaton concatenate(List<Automaton> l) {
     }
 
     result.finishState();
-
-    return result;
+    return Operations.removeDeadStates(result);
   }
 
   /**