I know that this has been a question asked here before, but I'm still struggling to understand the concept. V.I. Arnold states that for "the configuration manifold $V$, the 'generalised momentum '$\textbf{p}=\dfrac{\partial L}{\partial \dot{\textbf{q}}}$ is a cotangent vector. I'm struggling to see how this works, a co-vector, $dx_i$, at a point $p\in M$ is defined as
$$dx_i:T_PM\rightarrow \mathbb{R}$$ $$dx_i(\textbf{v})=v_i$$ Where $\textbf{v}\in T_pM$. So a co-vector should map $\textbf{v}$ to the $i$'th component of $\textbf{v}$. How would this work for generalised momentum, since it would mean $$\textbf{p}(\textbf{v})=\dfrac{\partial L}{\partial \dot{\textbf{q}}}(\textbf{v})=v_i$$ But I don't see how this is the case.
You need to decide if your position coordinates are $x$ or $q$, you are using both. Per the last formula, keep using $q$.
As a 1-form $\mathbf p$ is $\sum_i p_i\,dq^i$, so that the evaluation $\mathbf p(\mathbf v)$ has the result $\sum_i p_iv^i$.
Note that as directional derivative $$\frac{∂L}{∂\dot{\mathbf q}}(\mathbf v)=\frac{d}{ds}L(\mathbf q,\dot{\mathbf q}+s\mathbf v)\Big|_{s=0}$$ is indeed a linear functional in $\mathbf v$.