Some bikes have a "cutoff" signal wire you can cut which is a cheap option. If you're researching your bike and find "don't cut this wire" photo, consider that you shouldn't cut any wires and ask why they singled this one out.

You can likely find the controller in the frame and if you cannot find a guide for modifications otherwise, swapping that out for one that's from China or otherwise unrestricted should be a solid way to get the best out of the motor and all.

I'm fat, like very, especially by EU standards, so I added a second motor and controller to mine. Both have a thumb throttle so I can "go" without pedaling when I turn that option on in software.

Most controllers are universal for all countries so they can usually be configured for all sorts of things. Find the model number of your controller first.

Aside from wires that can be cut there could also be a sequence of steps you can do with the display buttons and thumb controller to allow thumb control only.

Aside from that, if you can get into the controller settings you can change the size of the wheels. Smaller wheels will make the walk assist faster but it will reduce the top governed speed of the bike and make the speedo read wrong, but it could be a quick fix.

I know how to hack any of them externally to alter the top speed, but I think you'll need a more advanced hack to pull off what you want to do here. I likely could still do it externally without altering the OEM controller circuit, but I would need some more advanced electronics to pull it off.

Both the speed and pedal cadence sensors are likely just a magnet passing in front of a hall sensor. It is possible to add anything that will trick this sensor into thinking there is a magnet passing in front of it.

You could make or 3d print an attachment to the crank arm (or wherever the pedal cadence sensor is located) - that has an additional magnet or few. This would cause the computer to think each pedal stroke is n × magnet strokes thereby reducing the ratio of effort to output all the time.

If you only want the effect to alter idle operations, it would need to be a microcontroller that detects the conditions and pulses something like an wire coiled inductor on a powder core ferrite bobbin placed near the hall effect sensor. You would need a sensor to detect the same pulse from the magnet on the crank arm and another on the wheel to detect rotation.

The code is not that complicated really, but debugging and eliminating potential noise sources would be pretty dangerous to prototype.

This is all overkill hypothesis and rambling to hopefully help you see how the system works in general on a fundamental level. For most people, a mechanical hack like additional magnets is pushing their maker skills pretty hard.

It is unlikely you are the first to have this issue and there is probably a much easier way. I'm interested in what others may say, so I commented to remember to watch this.