feat(store): enable deep comparison of objects in diffs

[LoneRifle]

When determining diffs, old and new values for a given row of data are compared using the equality operator. This does not work for objects, which would need comparison of the entire object. Further, the store currently assumes objects to be strings when generating schema.

A possible solution would be to recognise object values during schema generation, and performing a deep comparison for such values when determining diffs.

This PR consists of two parts:

refactor(store): separate number and string-related schema generation

• Move number-related metric type identification into one contiguous block; if the value is a finite number, return early with either a year or number type, depending on the name of the metric
• Similarly, move the identification of string-related types into its own block at the end; give the type as either a string or category depending on the number of unique values for the metric

feat(store): enable deep comparison of objects in diffs

• Identify objects during schema generation
• Use lodash's isEqual() to compare values if cell type is object

[Wattenberger]

looks great! thanks for digging into that. I added support for object cell types, which provides a little peak at the underlying data