Additional axes can inherit lazy eval'd ticks

[jiwalker-usgs]

As in this example

x <- seq(as.Date("2013-01-22"), as.Date("2013-10-02"), "days")
y <- rnorm(length(x), 71, 19)
gs <- gsplot() %>%
  points(x, y) %>%
  date_axis(side=1, pos.lab="interval", tick.int="month", snap.to="month") %>%
  axis(side=c(3,4), labels = FALSE) %>%
  grid()
gs

The ticks on side 3 should match up with side 1, but currently the ticks are determined at print-time, so side 3 inherits the auto ticks for the predetermined limits.

[lindsayplatt]

Only happening w/ date_axis. Saving for later since it doesn't impact repgen