We need to show that:
$$d'(A,C)\le d'(A,B)+d'(B,C)\le\delta(A,B)+\delta(B,C).$$
Reversing the roles of $A$ and $C$ shows the other inequality in order to show the full triangle inequality for $\delta$. Now let's unravel all the suprema and infima.
- Goal: $d'(A,C)\le d'(A,B)+d'(B,C)$
- Let $a\in A$, goal: $d(a,C)\le d'(A,B)+d'(B,C)$
- $d(a,B)\le d'(A,B)$, thus goal: $d(a,C)\le d(a,B)+d'(B,C)$
- Let $b\in B$, goal: $d(a,C)\le d(a,b)+d'(B,C)$
- $d(b,C)\le d'(B,C)$, thus goal: $d(a,C)\le d(a,b)+d(b,C)$
- Let $c\in C$, goal: $d(a,C)\le d(a,b)+d(b,c)$
- $d(a,C)\le d(a,c)\le d(a,b)+d(b,c)$ and we're done.
So actually there is no completeness necessary for verifying the triangle inequality - just follow your nose on the definitions and it proves itself. The above is a "backwards proof" along the lines that one would normally approach the solution; it can be rewritten in forward proving style as follows:
Let $a\in A,b\in B,c\in C$. Then $d(a,C)\le d(a,c)$, and $d(a,c)\le d(a,b)+d(b,c)$, so by $d(b,C)=\inf_{c\in C}d(b,c)$ we have $d(a,C)\le d(a,b)+d(b,C)\le d(a,b)+d'(B,C)$. Since this is true for every $b$ we have $d(a,C)\le d(a,B)+d'(B,C)\le d'(A,B)+d'(B,C)$, and since this is true for every $a$ we have $d'(A,C)\le d'(A,B)+d'(B,C)$ as desired.