(Short answer below) A phase diagram like the one you posted is to be intended true for a solution(either solid or liquid) of the two components(i.e. Al and Cu). Adding Cu to molten Al does not properly mean to have a solution(the two components must be dispersed in thw whole bath in a uniform way). Also a phase diagram is drawn as a "succession of equilibrium state", thermodynamic gergon for "waiting a long long time between any temperature or concentration variation".
This means that if you put Cu pellet in a molten Al bath the following happens: slowly the outer shell of copper will dissolve into Al (as tey are soluble as the diagram shows at the 30% wt%), at a certain point all the Cu will be melt in the bath. How long will it take? It is difficult to say(it can be computed though), depends on diffusion coefficient, temperature, stirring, ...
What is done for a practical application: Al and Cu are melt at about 1100°C (melting point of Cu) and then cooled down slowly below 582°C. When I say slowly I mean really slowly in order for the proper grain structure of the eutectic to be reached and to be uniform. I am not sure for the specific case but rather often this process has to be carried out in a controlled temperature environment for the cooling to be slow enough, this means atmospheric air cooling is often non viable.
From a thermodynamic point of view adding the copper pellets will cause them to melt even at the lower temperature but from an engineering/practical point of view it will take too long and material properties might not be the desidered ones.