The problem with this question is that static friction and kinetic friction are not fundamental forces in any way-- they're purely phenomenological names used to explain observed behavior. "Static friction" is a term we use to describe the observed fact that it usually takes more force to set an object into motion than it takes to keep it moving once you've got it started.
So, with that in mind, ask yourself how you could measure the relative sizes of static and kinetic friction. If the coefficient of static friction is greater than the coefficient of kinetic friction, this is an easy thing to do: once you overcome the static friction, the frictional force drops. So, you pull on an object with a force sensor, and measure the maximum force required before it gets moving, then once it's in motion, the frictional force decreases, and you measure how much force you need to apply to maintain a constant velocity.
What would it mean to have kinetic friction be greater than static friction? Well, it would mean that the force required to keep an object in motion would be greater than the force required to start it in motion. Which would require the force to go up at the instant the object started moving. But that doesn't make any sense, experimentally-- what you would see in that case is just that the force would increase up to the level required to keep the object in motion, as if the coefficients of static and kinetic friction were exactly equal.
So, common sense tells us that the coefficient of static friction can never be less than the coefficient of kinetic friction. Having greater kinetic than static friction just doesn't make any sense in terms of the phenomena being described.
(As an aside, the static/kinetic coefficient model is actually pretty lousy. It works as a way to set up problems forcing students to deal with the vector nature of forces, and allows some simple qualitative explanations of observed phenomena, but if you have ever tried to devise a lab doing quantitative measurements of friction, it's a mess.)
I am not sure why you are rejecting the static friction on the basis on long the parts are in contact. A "bond" is not a chemical bond that might take time, but rather an interaction between adjacent molecules, or atoms. It propagates at the speed of light, so there is plenty of time for the adjacent molecules to "bond" when sufficiently close enough.
In real life though, pairing down the tire/road contact into a friction coefficient is the wrong approach. It is a non-linear contact, where the higher the normal load the wider the contact patch is and the distribution of contact pressures changes. In addition, some parts have micro sliding as only 1 point in the contact patch is truly stationary.
There is something called the "Pacejka Magic Formula" which is a well established model of a tire contact, and there are newer ones out there which minor and major refinements to it. In the end, it depends on what you want out of it, in order for you to decide what contact/traction model to use.
[ref: Magic Formula]
Best Answer
The Laws of Dry/Solid Friction - such as that the friction force is independent of area of contact, proportional to the normal force between surfaces, and is independent of their relative speed - are empirical laws based on observation, like Hooke's Law and Ohm's Law. They are generally and approximately true in most situations for relatively small forces and speeds. Various theories are put forward to explain them, such as the making and breaking of bonds ('stick/slip'). But that is only one of several processes which are involved. Real surfaces also have small amounts of contaminants such as oil or grease which provide lubrication and complicate the situation.
Like all empirical laws they break down in more extreme situations or when tested with greater accuracy. As the speed of relative motion increases the friction force does in many cases decrease as you suggest it should, because of heating or aging.
We assume the Laws of Friction are true because they are useful, convenient approximations. They enable us to calculate and predict approximate analytical results without too much effort, especially when our object is to learn how to develop and use models to solve problems, rather than to model a real situation accurately. When accuracy does matter, more realistic empirical laws are used instead, and because of the complexity of these laws numerical calculations have to be made.
https://hal.archives-ouvertes.fr/ensl-00589509/document - courtesy of AnnaV https://arxiv.org/abs/cond-mat/0506657