Refactor REPS to use episode batch

[yeukfu]

Partially address https://github.com/rlworkgroup/garage/issues/1110. @ryanjulian @krzentner It seems that the implementation of REPS doesn't use baseline predictions, and I don't see that as input of tensorflow. Should we remove the bassline from its arguments?

[codecov[bot]]

Codecov Report

Merging #2123 into master will increase coverage by 0.01%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2123      +/-   ##
==========================================
+ Coverage   93.16%   93.18%   +0.01%     
==========================================
  Files         194      194              
  Lines       10525    10526       +1     
  Branches     1323     1321       -2     
==========================================
+ Hits         9806     9809       +3     
+ Misses        474      472       -2     
  Partials      245      245              

Impacted Files	Coverage Δ
src/garage/tf/algos/reps.py	98.46% <100.00%> (+1.03%)	:arrow_up:
src/garage/np/_functions.py	84.42% <0.00%> (-1.64%)	:arrow_down:
src/garage/sampler/multiprocessing_sampler.py	90.90% <0.00%> (+1.29%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 18f4a9b...b916c93. Read the comment docs.

[yeukfu]

@Mergifyio rebase

[mergify[bot]]

Command rebase: success

Branch has been successfully rebased

[yeukfu]

@Mergifyio rebase

[mergify[bot]]

Command rebase: success

Branch has been successfully rebased