Update tracking deception to output in TACC log

[wwongkamjan]

I changed followings in Cicero and prototxt so that Cicero would not filter messages that are not corresponding to its initial pseudo orders (PO):

1. in mila_api.py: I call msg['deceptive'] to log on TACC if there this bot is deceptive
2. in searchbot_agent.py: I make sure that we call a correspond function within a generate message function to check if the message corresponding to PO
3. in message_edit.py: I changed the line so that it won't return True immediately since it has no corresponding PO threshold but instead assign the default threshold and make sure that it later returns True with extra information about pseudo correspondence prob
4. I also copy contents in cicero.prototxt to create cicero_lie.prototxt. The difference is just remove line 66 pseudo_orders_correspondence_threshold: -5e-3 from cicero_lie.prototxt.

Also, these are commands that I run.

module load tacc-singularity
export WORK=/work/08801/wwongkam/ls6/ALLAN
export CICERO=/work2/08801/wwongkam/ls6/CICERO/cicero


cd $CICERO
singularity run  --nv \
  --bind "$WORK"/diplomacy_cicero/fairdiplomacy/agents/:/diplomacy_cicero/fairdiplomacy/agents \
  --bind "$WORK"/diplomacy_cicero/fairdiplomacy_external:/diplomacy_cicero/fairdiplomacy_external \
  --bind "$WORK"/diplomacy_cicero/parlai_diplomacy:/diplomacy_cicero/parlai_diplomacy \
  --bind "$WORK"/diplomacy_cicero/fairdiplomacy/AMR/:/diplomacy_cicero/fairdiplomacy/AMR/ \
  --bind "$CICERO"/agents:/diplomacy_cicero/conf/common/agents \
  --bind "$CICERO"/models:/diplomacy_cicero/models \
  --bind "$CICERO"/gpt2:/usr/local/lib/python3.7/site-packages/data/gpt2 \
  --pwd /diplomacy_cicero cicero_latest.sif

python fairdiplomacy_external/mila_api.py --game_type 2 --deceptive --game_id three_cicero_wwongkam --host shade-dev.tacc.utexas.edu --power RUSSIA --outdir $WORK/diplomacy_cicero/fairdiplomacy_external/out

The problem that I am having right now is that with changes in searchbot_agent.py line 1369 is not printing anything.