Improve autoSkip performance

[benmccann]

No description provided.

[kurkle]

Is the order of ticks lost? Especially when majors are enabled.

[benmccann]

Good catch @kurkle! I've updated it to address that

[kurkle]

Wondering why its not catched by a test. I don't get your fix through.
Aren't all major ticks still going to be first in the array, followed by the minors?
Separating minor/major would solve it without sorting, but would be a big change.

[benmccann]

Wondering why its not catched by a test.

Good question. There's a test in scale.time.tests called should handle autoskip with major and minor ticks that looks like it should catch it. Perhaps the assert library doesn't care about array order? There's no user-visible difference if ticks are plotted out-of-order, but I agree we should keep them in order anyway

I don't get your fix through. Aren't all major ticks still going to be first in the array, followed by the minors?

I don't think so. There's three cases in autoSkip:

1. A subset of major ticks are displayed
2. All major ticks and some minor ticks are displayed
3. Only minor ticks are displayed

For 2, which it sounds like you're most worried about, skip is called once for each set of adjacent major ticks and it will add the major tick and following minor ticks. It doesn't operate on all major ticks first