The event that occurred was a rapid onset issue. A couple of potential causes of a rapid lateral departure at speed are a failure of a slat track; (query Flash 604 Sharm al-Sheik) a high-speed turbulence encounter resulting in rapid alternating rudder inputs and consequent structural failure, a T/R deployment that is not recognized or responded to (Lauda 4); loss of IRU's, an attitude/erroneous attitude on one side not identified (a roll channel only failure can end badly... EGSS, B742F KAL 8509, VABB, AI855... Adam Air DHI574... (for own goals)
The B735 is quite a different autopilot architecture to the NG/MAX; a trim runaway is possible, as it is on any aircraft, but as had been discussed in depth following the Max events, a runaway in the first instance can be countered by control column input or the pickle switches and most appropriately by the trim cutout switches just behind the throttles. The aircraft has the same inherent issues if way out of trim, but it is an unlikely scenario for the plane given the events of the last 2 years.
If the mechanical reasons for a problem don't show evidence, then the human causes become of more concern. The parallels to MI185 Silkair are matched by the number of differences, and those suggest it is unlikely to be another in the sad list of intentional losses by a crew member. Simply, the time and height that the event occurred at is inconsistent with historical cases, where the pilot that causes the problem is sitting by themselves for a period of time. at 10K, both pilots will still have been in their seats. Intervention by the 2nd pilot will almost certainly occur, and that essentially is contrary to all of the historical cases.
I would be discounting a pitot-static problem in this case; while they occur with monotonous regularity, they almost invariably result in vertical deviations as well as IAS changes, but in this case, the ADSB data suggests otherwise. Dependent on setup and system selected, the ADSB data recorded may be the same or different from that shown on the MASI/ALT etc... The DFDR picks up the signal from the DFDAU which also converts the data sentence between 717 and 429 (don't have my refs with me... and its a few years since I was doing flight test on that aircraft... ) The data however will be able to be reconstructed from the known endpoint, and errors in P-S data compared to the necessary kinematics. The first glance suggests that this is a rollover event to the left, resulting in a very low nose attitude within a short period of time. That suggests attitude indicator issue/slat/T/R or unrecognized engine failure/rudder hard over or something similar. All of those have historical precedents, and all are able to be recovered from by an alert crew. COVID 19 has not left the industry in the best of shape... The CVR will give considerable information, some information will be available from the DFDR if recovered. This aircraft should have control input sensors being recorded, but limited flight control section position would be recorded in the -500 if my memory does not fail me.
Don't be too fast to blame the flight crew for active causation, this looks already like some event that was not able to be responded to effectively.
IMHO