highestVisibleXIndex has regression issue

[liuxuan30]

@danielgindi
for 375d7ba, you mentioned this closes for #785, #794

However, my old PR: 862de89 was trying to fix other issue.

I think we need review these two commits for highestVisibleXIndex? I had regression on my side:

pt.x now is 25.0, if we print it, but in the mem, it looks like 24.999999993, so floor(pt.x) would be 24, while it should be 25.0?

[liuxuan30]

So I checked #785, #794, I think they are ONLY related to lowestVisibleXIndex, because old code has "+1" appended. The latest code removed +1.
However, I think highestVisibleXIndex should stick to round as mentioned above.

[danielgindi]

Fixed in Charts 3.0 (v3 branch)

[liuxuan30]

Is it a fix or a new implementation so this is solved automatically? If so I will close #1000

[danielgindi]

The whole X concept changed, so it's a new implementation.
There's no "whole" x index, there are floating point x values, so we don't need to consider rounding to either side...

For x indexes there were reasons for both sides, used in multiple situations- now the situation does not exist :)