* Reindent TimeSeries classes
* Fix spacing in time series tests
* Remove trend tests where current is nil
I think if we've gotten this far with a nil value for current, there's a data integrity problem.
If we allow this, we'll have to be very defensive in our code. Best to raise and fix early.
* Reindent Money class
* Refactor TimeSeries artifacts
* Use as_json in TimeSeries
* Bring back tests for trends where current is nil
* Bring back trend test
* Correctly enumerate trend test
* Use favorable_direction for trend_styles helper
* Make trend public in TimeSeries::Value
* Allow nil current values in trends
I think I might've gotten it wrong before, nils might appear in trends if values are unavailable for snapshots
* Clean up TimeSeries::Trend
* Skip trend values same class validations if any values are nil
* Refactor Money
* Remove object parsing in TimeSeries::Value
We're only every passing hashes
* feat: add crypto account type to demo data
* fix: set currency to BTC, revert schema migration change and fix tests
* fix: update dates in time_series tests
* Clean up time series models
* Add value group rollup class for summarizing hierarchical data
* Integrate new classes
* Update UI to use new patterns
* Update D3 charts to expect new data format
* Clean up account model
* More cleanup
* Money improvements
* Use new money fields
* Remove invalid fixture data to avoid orphaned accountables
* Update time series to work better with collections
* Fix tests and UI bugs
