Refined vs unrefined anomaly?

In the constants section it says:

```{.python}
# Indicate which algorithm to use to calculate anomaly scores.
ANOMALY_DETECTOR_ALGORITHM = 'bitmap_detector'

# Indicate which algorithm to use to get refined maximal score within each anomaly.
ANOMALY_DETECTOR_REFINE_ALGORITHM = 'exp_avg_detector'
```

What does this mean?

Start at the constants section containing ANOMALY_DETECTOR_ALGORITHM and ANOMALY_DETECTOR_REFINE_ALGORITHM, then trace the referenced anomaly-detection algorithms to understand their roles. Done means the documentation clearly explains the difference between refined and unrefined anomaly scores and what each setting controls.

Written by the indexing model from the issue text.