It appears in the early portion of the video that pointing directly at the window is more sensitive than after 0:13 where you started pointing forward - probably the water drops, finger prints and dirty window are causing it to hunt.

With my phone 'tap and hold' sets AF and locks it at that point.

The OnePlus website says that your cellphone, the OnePlus 3T, uses a Samsung 3P8SP sensor (16MP, 1.12 μm ISOCELL) which uses on-sensor PDAF. This type of autofocus works on moving objects, at a distance, but requires enough light. Your conditions are a bit dim but the objects on the window, and occasionally the window frame at your camera's frame edges, are what's causing the problem.

Your phone has manual mode which allows control of the focus, lock it so distant objects are either in perfect focus or slightly off (allowing for nearer objects to maintain focus).

With such a tiny sensor the depth of field in focus ought to be very large, allowing all but closest and extremely distant objects to be excluded from a common focus setting (you might need to choose between extremely near or extremely far but if you exclude one of those two factors then everything ought to be sharply focusable).

You probably want to pick a clean spot on the window and shoot between the drops instead of having them bouncing around and making the video look so poorly stabilized (when distant objects are in fact fairly well stabilized, considering how bouncey the ride seemed). Move the camera closer to the window. Wrap your fingers around the phone (obviously not blocking the sensor) and let your nails slide on the glass as you bounce, all while tilting at the forward angle and holding level.

It would be great if the bus driver cleaned their own windows before taking tourists since people probably want to shoot photos and videos as they go. Find out it you can see the bus before you book a ride, if it's just public transit then you probably take what you get. Carrying a Windex® dampened lint-free cloth in a Zip-Lock® will make you look like an obsessive and weird person, whom probably will not have people sit next to them, and a further bonus of clean shots.

Inquire if they stop regularly for 5 minutes at points along the way to the destination so you can get out to shoot some stable and clean video.

The window was probably the culprit, and not locking the AF, it's certainly the bouncing water drops and dirty window that were the worst (appearance wise).

All-in-all it's 'OK' if you took a short clip of that video, where it worked out best, and made the majority of your video composed from outside the bus; making the majority a bumpy and poorly focused bus ride made for a seasick adventure.