Bearfoil, I remember calculating or at least generating a good estimate of the "beamwidth" for TCAS. With no satellite tracking a 45 degree or so banking turn to the left would break the beam. A heavy bank would mean TCAS could not maintain track, even though it has some beam steering capabilities. The fuselage would shadow it.
This question and mm43's observation about automation really starts me wondering if something happened to the crew way back shortly after the frequency/selcal switch. If they were out of action and the AP kicked out what would the plane do in severe turbulence?
(But, there are indications that between the pickle point and the ocean surface the plane's attitude did not switch such as to start dumping stuff on the floor such as a severe roll without a turn. This idea is out there to be shot down. {^_-} It helps build the event trees.)
I believe the VS does include the HF antennas. I believe VHF is on the fuselage and would still be present. (Experts know better than I. I've misplaced the antenna position plots that have been posted here.) And 121.5 is supposed to be continuously monitored by everybody in the air and on the ground concerned with aircraft in flight operations. Of course, even with the VS present they were effectively out of HF communications. No SWL, Short Wave Listener, has come forward with intercepts of their attempts to communicate, so far. And I don't know of DAKAR maintains days long recording logs BEFORE the selcal squelch of everything received on all their assigned frequencies. I doubt it. If there was such a log it might be illuminating. Otherwise HF is out of the picture here. (And if they called nobody would have heard - everybody else would have had their own selcal on for noise reduction.)
With active pilots present anecdotes indicate control is possible. Severe banking turns may become "an adventure", however. (correct me if I am wrong) If that is the case the plane should have been found since it would have flown at least far enough to end up in the areas already searched, I believe. That is probably why they were searched. It's not reasonable to presume the plane turned back either voluntarily or through some series of misfortunes.
The chevron shape? Well, wind sheer exists. Why not ocean current sheer? That is one hypothesis.
Regarding the IRUs I'm dragging out some old GPS knowledge here. Position is known relatively precisely depending on the GPS equipment involved. GPS alone can provide good estimates of long term velocity and crude to very crude estimates of attitude, velocity, and acceleration over short intervals. If the position error is say 5 meters and the sensors are 100 meters apart you can calculate for yourself the likely attitude errors from GPS alone. Also note that the position is not that if the receiver. It is the position of the antenna. A GPS receiver on a 100' cable to an active antenna can demonstrate this. Move the receiver around relative to the antenna. The position does not change by 200 feet or even 30 feet as you move around the extreme perimeter allowed by the cable.
This attitude issue is the part of the navigation solution the IRUs are designed to provide. And it is known that their sensor positions affect their data relative to the vehicle. So that is entered into the calculations (at least on military sets I know about.)
Where my knowledge leaves off is the dynamic range of the various IRU sensors when subjected to sudden high accelerations. If a sensor bumps its limits it's suddenly useless. Also note that in the dark at night with no radar to find a real horizon there is no way to distinguish gravitational acceleration from any other acceleration. So you probably lose all sense of precise attitude if a sensor hits its limits. So an intact aircraft and healthy aware crew can suddenly find themselves in the terrifying position of flying blind without even reliable seat of the pants observations if too many of the old crude instruments have been eliminated.
{^_^}