add new "justify" scale type

[leeoniya]

there are several viz types that can benefit from a built-in scale that is ordinal with respect to the number of datapoints and/or number of visible series.

grouped bars demo does its own scale math internally to distribute the bars correctly. it does a 2-level distribution -- first groups are spread based on number of datapoints and then then bars within those groups are spread based on number of series. this is done along x axis: https://leeoniya.github.io/uPlot/demos/multi-bars.html

state timeline and status history distribute lanes along y axis based on number of series. the y "values" represent the color, so it's pretty weird: https://leeoniya.github.io/uPlot/demos/timeline-discrete.html

it's time to internalize at least the scale math properly. dealing with these bits externally is a huge headache.

• the scale settings need to account for both spacing and sizing along series and/or values counts. it will follow the css justify-content distributed alignment:

  /* Distributed alignment */
  justify-content: space-between; /* Distribute items evenly
                                   The first item is flush with the start,
                                   the last is flush with the end */
  justify-content: space-around; /* Distribute items evenly
                                   Start and end gaps are half the size of the space
                                   between each item */
  justify-content: space-evenly; /* Distribute items evenly
                                   Start, in-between, and end gaps have equal sizes */

• this scale type will likely not be zoom-able. the scale.range function will be force-set to an appropriate calc function which allows integer index positions to land at the correct pixels. there will need to be a new api for these scales that accepts series index, value index, and whether to return left, center, or right positions of an item.
• series visibility toggle will need to internally re-distribute these scales and be able to skip the disabled series in the apis while maintaining index associativity.
• do something sensible for the axes tick labels

this is a significant undertaking that will likely touch 70% of the codebase. with many unknowns still...