Quantum mechanical entities are described by solutions of the Schrodinger equation, the wave function, with specific boundary conditions. A measurement changes the boundary conditions and thus the subsequent wavefunction describing the particle measured will be a different one.
A measurement picks an instant from the probability distribution describing the entity, and this is called "collapse" to confuse people. It is one entry in a distribution
Lets take the simple example of the harmonic oscillator:

On the right is the square of the wave function, and one sees oscilations in the probability distribution in the displacement x.
Suppose we have one atom the electron's dependence described by a harmonic oscillator . If we measure the displacement, it means we probed the atom and we will get one value in the distribution. A second atom will give a second point etc, and we will develop histograms that will have the shapes on the right. Each atom-electron after the measurement will be described by different wavefunctions because the boundary conditions will have changed, for example "scattered out" , "on a different energy level, etc. It will be very hard to follow what happens collectively from then on to determine the new probability functions because the boundary conditions will depend on the scatter.
Measurements of elementary particles, quantum mechanical entities, usually manifest the particle side . In this electron double slit one at a time one sees the particle

Electron buildup over time
as it leaves a spot on the screen, and each spot with its relevant probability looks random. The accumulation of spots though shows that the probability of finding the screen has a wave nature. This is analogous to the example above. When the electron hits the screen the problem changes because the boundary conditions change due to the screen
Ex. if a particle was measured before the slits, would we see an interference pattern, or a particle pattern?
It will depend on the boundary conditions of the measurements before. If it is a screen it is self evident . If the experiment is set up that the boundary conditions after the first measure allow again a coherent wavefunction plot with the boundary conditions of "two slits electron" the experiment will be the same as in the picture. There have been experiments interfering with the incoming beam statistically and maybe you can understand them.
And if the wave function is physically collapsed does it say like that forever or does it ever revert back to it's "superposition" state.
It always hinges on boundary conditions of the experiment changing the wavefunction solutions relevant to the experiment.