Improve checkpoint figure and table generation

[AdamGleave]

• Bugfix: in rollout_return, have lower CI correspond to higher return (so lower implied "distance") and upper CI correspond to lower return.
• Checkpoint timeseries: generate one figure per algorithm (for all rewards) and per reward (for all algorithms). Various changes to pretty-print: notably including separate x-axis for return; qualitative colour schemes (differing between algorithm and reward scheme);

[codecov[bot]]

Codecov Report

Merging #51 (70abf54) into master (b8cdfc9) will decrease coverage by 1.25%. The diff coverage is 33.33%.

@@            Coverage Diff             @@
##           master      #51      +/-   ##
==========================================
- Coverage   82.86%   81.60%   -1.26%     
==========================================
  Files          69       69              
  Lines        4796     4926     +130     
==========================================
+ Hits         3974     4020      +46     
- Misses        822      906      +84     

Impacted Files	Coverage Δ
src/evaluating_rewards/analysis/stylesheets.py	69.56% <ø> (ø)
...ing_rewards/scripts/pipeline/combined_distances.py	54.41% <15.32%> (-12.05%)	:arrow_down:
src/evaluating_rewards/distances/common_config.py	92.95% <100.00%> (+2.57%)	:arrow_up:
src/evaluating_rewards/scripts/distances/common.py	90.20% <100.00%> (+0.28%)	:arrow_up:
...uating_rewards/scripts/distances/rollout_return.py	97.14% <100.00%> (+0.31%)	:arrow_up:
src/evaluating_rewards/scripts/rl_common.py	80.00% <100.00%> (ø)

---

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 b8cdfc9...70abf54. Read the comment docs.