Facilitate error analysis when using `tongfen_estimate`

[mountainMath]

tongfen_estimate works for all combinations of source and target geometries, but it is only an estimate and will produce errors. To be useful for research we need effective ways to understand the effects of those errors on subsequent analysis.

A possible solution is to generate multiple estimates that reflect the range of errors for each target geography. However, it is not clear how to encode the very real possibility of correlated errors between neighbouring geographic regions. Someone must have though about this before ...