Using two Y axes and a Y axis grid easily leads to the situation, that the grid
lines don't match the interval ticks of the secondary Y axis - because they
originate from the primary Y axis.

This makes it very hard to properly read values for data series attached to the
secondary Y axis.

If the two axes were aligned, i.e. if they had the same number of intervals, the
grid would match both axes, and data would be easier to read.

There are two ways to achieve this: leave the scale of the secondary Y axis as
is and just adjust the number of intervals on it. This isn't too smart, because
one may end up with secondary Y axis intervals at 16.4, 32.8, 49.2 etc.

The second, and probably better way is to adjust the scale of the secondary Y
axis, by increasing it's maximum (or minimum, in case of negative numbers) to
some smart value, so that automatically calculated intervals would match the
ones from the primary Y axis. With "smart" I mean something like "to the next
full number, full thousand, full million" or whatever matches the scale of the
axis best.

Easy example: 
- Two data series: (1, 2, 3) and (3, 4, 5). 
- Map second to secondary Y axis.
-> Primary Y axis is scaled from 0 to 3.5 (seven intervals)
-> Secondary Y axis is scaled from 0 to 6 (six intervals)
If the second Y axis had it's maximum at 7, it would have had the same number of
intervals, ad a Y axis grid would work for both.