Fixed code extraction issues for bigLength and bigSubstring

Fixed compilation issue in String.scala
Made sure (...).substring(start) is extracted as a lset in order to not recompute the body twice (and adapted the pretty-printer)