What's Changed
* Update priors and other updates by thodson-usgs in https://github.com/thodson-usgs/ratingcurve/pull/19, including
* Revert to pymc5
* Revert to sorted breakpoints
* Fix uncertainty
After lengthy testing, pymc5 seems to give comparable performance to pymc4. Sorting the breakpoints helped diminish the discrepancy between versions, but some of the difference is still unexplained.
**Full Changelog**: https://github.com/thodson-usgs/ratingcurve/compare/v0.3.3...v0.3.4