Consider an example of a circular trajectory:
$$\mathbf{r}(t)=\cos(t)\hat{\mathbf{x}}+\sin(t)\hat{\mathbf{y}}$$
We can rewrite this purely in terms of cosines, by remembering that $\sin(t)=\cos(t-\pi/2)$:
$$\mathbf{r}(t)=\cos(t)\hat{\mathbf{x}}+\cos\left(t-\frac{\pi}{2}\right)\hat{\mathbf{y}}$$
This corresponds to counterclockwise motion. If we take $t\to-t$, then clearly we will be left with clockwise motion:
$$\begin{align*}\mathbf{s}(t)&=\cos(-t)\hat{\mathbf{x}}+\sin(-t)\hat{\mathbf{y}}\\\sin(-t)=-\sin(t)\implies&=\cos(-t)\hat{\mathbf{x}}-\sin(t)\hat{\mathbf{y}}\\\cos(-t)=\cos(t)\implies&=\cos(t)\hat{\mathbf{x}}-\sin(t)\hat{\mathbf{y}}\\&=\cos(t)\hat{\mathbf{x}}+\cos\left(t+\frac{\pi}{2}\right)\hat{\mathbf{y}}\end{align*}$$
You can see that the two forms are the same, except in the clockwise case the $\hat{\mathbf{y}}$ component has a phase shift which brings it "ahead" of the $\hat{\mathbf{x}}$ component, and vice versa for the clockwise case. This is what is meant by one component "leading" the other.
So, in your case for a general elliptical trajectory
$$\mathbf{r}(t)=A_1\cos(t+\delta_1)\hat{\mathbf{x}}+A_2\cos(t+\delta_2)\hat{\mathbf{y}}$$ with $\delta=\delta_2-\delta_1$, we can determine the directionality of the trajectory based on $\delta$. Based on our observations above, we can see that $\delta<0$ implies the trajectory will be counterclockwise, and for $\delta>0$ it will be clockwise. If we wrap that into the typical $[0,2\pi)$ interval, we can say that a clockwise trajectory will have $0<\delta<\pi$, and a counterclockwise one will have $\pi<\delta<2\pi$. A directionality is not really defined for $\delta=n\pi:n\in\mathbb{Z}$ because the trajectory collapses to a line.