There's a TV news interview out there of with 3rd party testimony from someone on the phone with a victim who was complaining right before the collapse about the pool patio area sinking. (Edit 6/26/21: I am referring to the video link posted by Jon43442, above, where the interviewee at 1:30 says 'She was on the balcony on 4th floor and she was telling her husband in a frantic way that "the pool was sinking, the ground was cracking," ... "The pool is sinking. The ground is shaking" ... "The building's shaking! The building's shaking" and then she screamed a screech and the phone went dead.')
It's actually pretty plausible that a patio slab failure could have been the initiating event here:
There is otherwise not a whole lot of reason why the collapsed portion of the building zone should have caused as large a portion of the patio to fail as it did. Water ingress through failed waterproofing in the patio area could have damaged rebar connecting columns to the slab leading to shear failure and a single initiating punch-through of a single column. Such a punch-through could overload neighboring columns with similar damage leading to a cascading collapse in the patio zone.
The suspended patio would now have very high tensile stresses in its rebar with the rebar acting like the cables in a suspension bridge. The rebar on the bottom (east-west oriented) seems to have torn out relatively easily, relieving those stresses and potentially saving the southwest corner of the structure and making the load be carried in the north-south oriented rebar.
The extreme tensile stresses in the north-south oriented rebar from the collapsing patio would mean that the patio level slab, instead of providing lateral stabilization to the columns in the portion of the building that failed first, would instead be destabilizing those columns.
From the perspective of the patio slab, the (relatively) stiff intact columns become stress concentrators for those tensile stresses. The columns under the building proper, probably bigger and better protected from water/corrosion than those under the patio, would be less likely to have punchout failures and more likely to transfer the tensile loads into the sides of the columns. Now we have highly loaded vertical columns with a significant sideload. It would only take one of these to buckle to initiate the major collapse. I am thinking of our undergraduate mechanics of materials lab where students put a rod in compression and then tap it from the side and it buckles instantly. With few internal walls, such a single column failure may have been enough to destabilize the larger structure.