SchemaUtil.getNameParts(...): Use US Locale and split on '/' and '_'

Always use US Locale when converting name parts to lower case. This
makes the behaviour independent of the default Locale.

In addition also split names on '/' and '_'. Splitting on these
characters makes sense for group names, especially since in future
groups are stored as refs/groups/<group-name> refs in NoteDb and
hierarchical group names will be encouraged.

Both these changes to SchemaUtil.getNameParts(...) require new schema
versions for all indexes since values for fields that use
SchemaUtil.getNameParts(...) may change. This is why we do both
modification in the same change, so that we must increase the index
schema versions only once.

Change-Id: I26f6940216058acd27e2f0dd38441a93fe5d7107
Signed-off-by: Edwin Kempin <ekempin@google.com>

gerrit-server/src/test/java/com/google/gerrit/server/index/SchemaUtilTest.java

@@ -15,6 +15,7 @@
 package com.google.gerrit.server.index;
 
 import static com.google.common.truth.Truth.assertThat;
+import static com.google.gerrit.server.index.SchemaUtil.getNameParts;
 import static com.google.gerrit.server.index.SchemaUtil.getPersonParts;
 import static com.google.gerrit.server.index.SchemaUtil.schema;
 
@@ -68,4 +69,11 @@ public class SchemaUtilTest extends GerritBaseTests {
             "ba-z@example.com", "ba-z", "ba", "z",
             "example.com", "example", "com");
   }
+
+  @Test
+  public void getNamePartsExtractsParts() {
+    assertThat(getNameParts("")).isEmpty();
+    assertThat(getNameParts("foO-bAr_Baz a.b@c/d"))
+        .containsExactly("foo", "bar", "baz", "a", "b", "c", "d");
+  }
 }