It's important to realize that binary stars form much differently than planets do. Assuming that both stars form in situ (i.e. excluding scenarios where one is captured from outside the system), there are several main ways for a binary star system to form from a molecular cloud. The most widely-accepted model at the moment is the fragmentation hypothesis, where the protostellar cloud splits during its collapse, forming two clouds which will each collapse into separate protostars.
Planets, on the other hand, form not from collapsing clouds but in protoplanetary disks around stars. The dynamics of these systems are substantially different, as there are interactions between the gas, dust, and protoplanets - perhaps a more complicated system than a collapsing gas cloud. We therefore see planets arise through different processes, such as gravitational instabilities within the disk or pebble accretion.
The paper being referenced found that the GJ 3512 system appears incompatible with the pebble accretion models the team tested - that is, rocky planetary cores being built up through collisions. In a low-mass star system, the cores should move inwards too quickly to accrete enough mass, and tweaking other initial conditions didn't help. That's what the BBC article focuses on.
What the BBC ignores is that the authors were able to reproduce the system by assuming that the planet(s) arise through gravitational instabilities within the protoplanetary disk, and this model is borne out by the simulations. It may not have been widely considered, but it still works. That's one of the key results to pay attention to.