Given a continuous scale, returns a rounding function that snaps values to the coarsest precision that still distinguishes neighboring pixels. For temporal scales, this means utc or time intervals; for numeric scales, “nice” values = 10^x * {1, 2, 5}. For non-linear scales (log, pow, symlog, etc.) we compute precision locally.

This is a separate branch needed by both the dataless brush and the dataless crosshair

This was referenced

Fil marked this pull request as ready for review