I've seen a definition of continuity on the resolution of an excersice that left me a little bit sceptical.
The definition stated that a function $f$ is continuous at a point $x$ if $\forall \epsilon>0$ there's a $\delta>0$ such that $|f(z)-f(y)|<\epsilon$ whenever $z, y \in B(x,\delta)$.
I may be ignoring something, but it seems that this definition is not ok. What about removable discontinuities? Any thoughts?
This is not the $\varepsilon-\delta$ definition of continuity in the context of metric spaces. However, it is equivalent to the usual definition, which is$$(\forall\varepsilon>0)(\exists\delta>0):d(x,y)<\delta\implies d\bigl(f(x),f(y)\bigr)<\varepsilon.\tag1$$In fact, if your condition holds, then just take $z=x$ abd $(1)$ will hold. And if $(1)$ holds, then, given $\varepsilon>0$, take $\delta>0$ such that $d(x,y)<\delta\implies d\bigl(f(x),f(y)\bigr)<\frac\varepsilon2$. Then the condition that you stated will hold.