Compute and report confidence intervals in heatmaps

[AdamGleave]

Compute confidence intervals using bootstrapping (all), and Student's t-distribution (CANON, EPIC). Also report sample mean and SD (CANON, EPIC) across seeds.

Visualized by saving multiple heatmaps, one for each type.

[codecov[bot]]

Codecov Report

Merging #30 into master will decrease coverage by 0.14%. The diff coverage is 88.82%.

@@            Coverage Diff             @@
##           master      #30      +/-   ##
==========================================
- Coverage   87.98%   87.84%   -0.15%     
==========================================
  Files          55       57       +2     
  Lines        3830     3957     +127     
==========================================
+ Hits         3370     3476     +106     
- Misses        460      481      +21     

Impacted Files	Coverage Δ
...s/dissimilarity_heatmaps/plot_gridworld_heatmap.py	92.30% <ø> (ø)
tests/test_canonical_sample.py	100.00% <ø> (ø)
tests/test_scripts.py	100.00% <ø> (ø)
...ysis/dissimilarity_heatmaps/plot_return_heatmap.py	92.10% <76.47%> (-5.08%)	:arrow_down:
...lysis/dissimilarity_heatmaps/plot_canon_heatmap.py	93.37% <78.57%> (-3.73%)	:arrow_down:
...alysis/dissimilarity_heatmaps/plot_epic_heatmap.py	93.40% <78.57%> (-2.89%)	:arrow_down:
...ards/analysis/dissimilarity_heatmaps/cli_common.py	77.09% <89.87%> (+9.13%)	:arrow_up:
...ewards/analysis/dissimilarity_heatmaps/heatmaps.py	90.69% <100.00%> (+0.10%)	:arrow_up:
src/evaluating_rewards/canonical_sample.py	97.84% <100.00%> (-0.27%)	:arrow_down:
src/evaluating_rewards/tabular.py	72.91% <100.00%> (ø)
... and 4 more

---

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 47156bc...6fb00a6. Read the comment docs.