LACIE_A2C outperform vanilla A2C

Scripts to rerun experiments

python main.py --num-stream-jobs 1000 --num-stream-jobs-factor 1.05\
                --num-curriculum-time 1 \
                --algo lacie_a2c \
                --num-env-steps 20000000\
                --gamma 0.97\
                --entropy-coef 0.01\
                --load-balance-service-rates 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 \
                --eval-interval 50\
                --reward-norm-factor 10000\
                --lr 0.0007\
                --num-mini-batch 32\
                --adapt-lr 1e-3\
                --num-inner-steps 5\
                --num-process 16 --num-steps 100 --log-interval 10 \
                --seed 26 --recurrent-policy --use-linear-lr-decay\
                --log-dir lacie_a2c_logs