Suporting unicode id's

Don't care about the actual content of id. HTML4 cares, but html5 does not. Even, in html4, is pointless manufacturing a new id to target that indeed it is not there. If the id is not legal searching for it will fail and the effect would be the current behavior.

It fixes  hakimel#2198. About pandoc generated ids based on unicode titles.

js/reveal.js

@@ -3691,12 +3691,8 @@
 		// assume that this is a named link
 		if( isNaN( parseInt( bits[0], 10 ) ) && name.length ) {
 			var element;
-
-			// Ensure the named link is a valid HTML ID attribute
-			if( /^[a-zA-Z][\w:.-]*$/.test( name ) ) {
-				// Find the slide with the specified ID
-				element = document.getElementById( name );
-			}
+			name = decodeURIComponent(name);
+			element = document.getElementById( name );
 
 			if( element ) {
 				// Find the position of the named slide and navigate to it
@@ -3743,9 +3739,6 @@
 
 				// Attempt to create a named link based on the slide's ID
 				var id = currentSlide.getAttribute( 'id' );
-				if( id ) {
-					id = id.replace( /[^a-zA-Z0-9\-\_\:\.]/g, '' );
-				}
 
 				// If the current slide has an ID, use that as a named link
 				if( typeof id === 'string' && id.length ) {