Spike: Morecast - Investigate options for bias corrected precip

EC raised a question surrounding bias corrected precip values. They asked if we can set the bias corrrected value to 0 when the model prediction is zero. There was some discussion during sprint review around this topic:
- one suggestion to get rid of bias adjusted precip
- another from MM saying we should keep it
- an option to always set bias adjusted precip to 0 when model prediction is 0
- the idea of separate handling during dry/wet periods

- one more comment from EC

"I agree the bias should be set to zero when the modelled precip = 0. I would also suggest that forecast precip < 0. 5 mm also default to zero bias as well. The threshold precip value for FFMC is 0.5 mm. Therefore anything less then 0.5 mm has no impact on fire behaviour. We would need consensus between forecasters to a"gree to this adjustment.

No files, tests, or entry points are named. Start by reviewing the listed alternatives and the unresolved forecaster-consensus question; done means selecting and documenting an agreed rule for bias-corrected precipitation, including the proposed 0.5 mm threshold.

Written by the indexing model from the issue text.