For each $g \in G$, we have the (smooth) right translation map $R_g: x \mapsto xg$. For each $h \in G$ this induces a (linear) derivative map $d(R_g)_h: T_h(G) \to T_{hg}(G)$. Given a tangent vector $v \in T_h(G)$, the authors define the tangent vector $vg \in T_{hg}(G)$ by pushing forward by this map:
$$
vg := d(R_g)_h(v) \in T_{hg}(G) \, .
$$
(One can show that this is indeed a right action on the tangent bundle $TG$ since $d(R_{g_1 g_2}) = d(R_{g_2}) \circ d(R_{g_1})$.) In particular, for $h = e$ and $v \in T_e(G)$, we get a tangent vector $vg \in T_g(G)$, which shows that $T_e(G) g \subseteq T_g(G)$.
Do you see how to show the reverse containment? Hint: Use $g^{-1}$.