2) MAX_FRAG_PER_NODE separated from NO_OF_FRAG_PER_NODE and
takes into account large fragmentation
3) SUMA use config parameter instead of hard coded limit


ndb/src/kernel/blocks/backup/BackupInit.cpp:
  Moved addition of system tables and its attributes to Configuration.cpp
ndb/src/kernel/blocks/dbdih/DbdihMain.cpp:
  Medium == default is one fragment per node
ndb/src/kernel/blocks/dbtux/Dbtux.hpp:
  Use MAX_FRAG_PER_NODE, NO_OF_FRAG_PER_NODE is used for
  allocation purposes, still allowed to create a table with
  many fragments
ndb/src/kernel/blocks/suma/SumaInit.cpp:
  Use max no of tables configured instead of maximum configurable
ndb/src/kernel/vm/Configuration.cpp:
  Update configuration to handle also system tables and their attributes
ndb/src/kernel/vm/pc.hpp:
  Maximum fragments per node is 8 in the worst case