Metrics: Default to numpy scalars for computation

[jarq6c]

External testing found that attempting to compute categorical metrics using contingency tables with zero-valued components resulted in a ZeroDivision error. This behavior was undesirable for downstream users and inconsistent with the behavior of other packages used by hydrotools. This update introduces conversion of contingency table components to numpy.float64 scalars. This has the advantage of inheriting all of the numpy built-in value checking, including raising a RuntimeWarning in the event of division by zero, instead of raising an error.

Additions

• convert_mapping_values is a new method used to convert dict-like objects to numpy DTypeLike values.

Removals

• None

Changes

• Categorical methods all start by calling convert_mapping_values on the input contingency table.

Testing

1. Explicit tests with null or zero-valued contingency table components
2. Adding test scenarios for null or zero-valued continuous metrics
3. convert_mapping_values is specifically tested using a pandas.Series

Notes

• None

Todos

• None

Checklist

• [x] PR has an informative and human-readable title
• [x] PR is well outlined and documented. See #12 for an example
• [x] Changes are limited to a single goal (no scope creep)
• [x] Code can be automatically merged (no conflicts)
• [x] Code follows project standards (see CONTRIBUTING.md)
• [x] Passes all existing automated tests
• [x] Any change in functionality is tested
• [x] New functions are documented (with a description, list of inputs, and expected output) using numpy docstring formatting
• [ ] Placeholder code is flagged / future todos are captured in comments
• [x] Reviewers requested with the Reviewers tool :arrow_right:

[jarq6c]

Related to an issue originally reported by @RossWolford-NOAA

[jarq6c]

Took me a sec to see what the before and after of convert_mapping_values was gonna be, but it seems pretty good to me.

Thanks for the review!