fix failure during start

1) move AbortOption from NdbTransaction to NdbOperation

2) let each operation have a "default" abort option dependant on
   operation type

   - read - AO_IgnoreError
   - dml - AbortOnError
   - scan take over - AbortOnError

3) Changed default value to execute() from AbortOnError to DefaultAbortOption, which does not change the operations abort-option.

   Another value to execute(AO) is equivalent to setting AO on each operation before calling execute

4) execute() does _only_ return -1 if transaction has been aborted
   otherwise, you need to check each operation for error code

  remove copy/paste method in NdbIndexOperation

  two small fixes for TC
    1) use static syntax
    2) fix attlen handling in unique index

  Fix read after delete (in same op)
  - lqh, handling of attrinfo
  - tup, setup read for disk/varchar

  - add some sanity check to marker/hash code to see that element isnt inserted twice into hashtable
    (if defined VM_TRACE or ERROR_INSERT)

  - allow REMOVE_MARKER_ORD to fail(dont find record) in release

  remove accidently left debug prinouts

  Make sure subscriber is removed from list when n_subscribers is decreased

  Fix assertion in corner-case where signal isnt sent to any node

  fix some more sp100 hang cases

into  perch.ndb.mysql.com:/home/jonas/src/51-work

  fix cpu peak in big clusters during unpack of config

into  perch.ndb.mysql.com:/home/jonas/src/51-work

  add support for doing mlockall before mallc instead of after
  (recommit in 5.0)

  additional fix after autotest
  dont send start_fragreq to temporary tables such as ordered indexes...

  Fix DD problem during NR after 3 missed LCP's

  fix not aligned or non 32-bit values in "smart" scan

into  perch.ndb.mysql.com:/home/jonas/src/51-work

  fix bug when master failure during nr
  (recommit against 5.0)

  handle partially transfered LCP_FRAG_REP after node failure
  recommit to 51-work

  extra sledge hammer, (real 51)
  use dicklock to prevent SUB_START_REQ during node recovery

  Fix most obvious error in dict/suma error handling of events

into  perch.ndb.mysql.com:/home/jonas/src/51-work

  on master node failure during qmgr-commitreq
    make sure to remove all committed failed nodes from failed/prepfailed arrays

into  perch.ndb.mysql.com:/home/jonas/src/51-work

  incorrect handling of commit/ignore error in unique index code

  add testcase

  make sure apicon timer is reset when setting CS_CONNECTED

  make sure not too poll to long, in case of hb-failure of last db-node

  make sure DISK flag is set on DATA column for blob table

  split assignment into 2 lines

  Fix in bug fix, make sure LCP_SKIP is only used for inserts...

  Fix start transaction with hint from ndbapi