Speed up split node

fixes #5251

The code changed from 4.0.x to 4.1.x

This change to to prevent making changes to the orginial input `msg`
object incase any values were stored in context (pass by refernce).

The change meant that for every output message the whole original
input `msg` was being cloned, which could be huge, causing a big
performance regresion.

This fix ensures the clone of the orginial `msg` is only done once
and the much smaller output message is then cloned again to update
`msg.parts` object for each output. This results in lots of small
clones rather than lots of very large clones.

packages/node_modules/@node-red/nodes/core/sequence/17-split.js

@@ -146,16 +146,16 @@ module.exports = function(RED) {
                     var pos = 0;
                     var data = value;
                     msg.parts.len = node.arraySplt;
+		    const newmsg = RED.utils.cloneMessge(msg)
                     for (var i=0; i<count; i++) {
                         var m = data.slice(pos,pos+node.arraySplt);
                         if (node.arraySplt === 1) {
                             m = m[0];
                         }
-                        const newmsg = RED.util.cloneMessage(msg)
                         RED.util.setMessageProperty(newmsg,node.property,m);
                         newmsg.parts.index = i;
                         pos += node.arraySplt;
-                        send(newmsg);
+                        send(RED.utils.cloneMessage(newmsg));
                     }
                     done();
                 }