bug in interpretedupdate (used for auto-increment) that made TUPKEYREF be sent twice
  this was however "normally" shadowed, but made LQH crash when LCP was about to start


mysql-test/r/ndb_basic.result:
  testcase
mysql-test/t/ndb_basic.test:
  testcase
ndb/src/kernel/blocks/dblqh/Dblqh.hpp:
  add extra variable to make sure TUPKEYREF is not received twice (VM_TRACE)
ndb/src/kernel/blocks/dblqh/DblqhMain.cpp:
  add extra variable to make sure TUPKEYREF is not received twice (VM_TRACE)
ndb/src/kernel/blocks/dbtup/DbtupExecQuery.cpp:
  make sure TUPKEYREF is not sent twice in case of "error" in interpreted program