Noncompact counterexample to the Šura-Bura lemma

Set $p=(0,0)$, $q=(1,0)$, ${\epsilon }_m=1/(m+2)$, and let $A_m$ be the polygonal arc joining $({\epsilon }_m,0)$, $({\epsilon }_m,m)$, $(1-{\epsilon }_m,m)$, and $(1-{\epsilon }_m,0)$. In the planar subspace $Z=\{p,q\}\cup {\bigcup }_m{A}_m$, each arc is clopen. The component of $p$ is therefore just $\{p\}$. Any clopen neighborhood of $p$ contains the left endpoints of all sufficiently high arcs, hence those entire connected arcs, and by closedness also the limit $q$ of their right endpoints. Thus its quasi-component is $\{p,q\}$. Higher arcs escape to infinity while their endpoints approach the two distinguished points.