Improve by-key reconciliation performance

[tferi]

Ordered reconciliation works fine in O(n), but the other reconciliation method has a higher complexity currently: https://github.com/tferi/templateFX/blob/7a9017cb9ba6e377fcbde6a9575a0b19b1eed5bb/base/src/main/scala/com/tothferenc/templateFX/collection/OrderedSpecsWithIds.scala

(Note that MoveNode is O(n) itself because an indexOf())