Generate and compare trie stress counts using M/R

[mikewalch]

See apache/fluo#277

[keith-turner]

The following is a rough procedure for determining where diffs occured.

Run following commands in Accumulo shell to create an empty fluo table with correct config.

clonetable fluo fluo_mr
deleterows -t fluo_mr -f

copy fluo.properties and change table to fluo_mr

point fluo stress env.sh to copy of fluo.properties

run following command to init fluo_mr with complete tree

./bin/init.sh /stress/*/* /stress/initTmp 9

Run following commands in accumulo shell to drop multiple versions, while keeping original table.

clonetable fluo fluo_clone
compact -t fluo_clone -w

Run following commands in bash to diff table at a level 7.

accumulo shell -u root -p secret -e 'scan -t fluo_mr -c count:seen -np -b 07 -e 08' | grep DATA | sed 's/[0-9]*-DATA/DATA/' > /tmp/mr-l7-2.txt 
accumulo shell -u root -p secret -e 'scan -t fluo_clone -c count:seen -np -b 07 -e 08' | grep DATA | sed 's/[0-9]*-DATA/DATA/' > /tmp/clone-l7-2.txt
diff /tmp/clone-l7-2.txt /tmp/mr-l7-2.txt