[FR] Allow user to specify higher-precision adapted_grid

[calumapplepie]

Add some function or keyword argument to allow for increasing adapted_grid curvature limits

[calumapplepie]

Its also possible that the case where I'm seeing this problem is algorithmic; I mostly see it with functions involving asymptotes at zero; eg, plot(x->x^-6,ylims=(0,10))

Perhaps its the interaction of the asymptote with the adaptive grid calculator that's the problem?