First you are correct, there is no fundamental difference in reactions being described as reversible or irreversible, unlike in thermodynamics. A reaction will be called irreversible (a)if the product is removed from reaction i.e. by precipitation or physical removal and (b) if the rate of returning from product to reactants is so slow that it cannot be measured. This is somewhat subjective, but the lifetime of the back reaction can be years, so in practice its clear cut.
Secondly, the equilibrium constant for a reaction $\ce{aA + bB <=>cC + dD} $ is defined as $$K_p=\frac{P_C^cP_D^d}{P_A^aP_B^b}$$ where $P_A^b$ etc. represents the pressure at equilibrium. As the equilibrium constant must be dimensionless so pressure is understood to be divided by unit pressure of 1 bar. If the reaction is in solution then concentrations are used instead of pressure, then these are understood to be divided by $\ce{1 mol~dm^{-3}}$ (Technically instead of pressure or concentrations activities are used but for most uses pressure and concentration are sufficient). The standard free energy is defined as$$ \Delta G^0 = -RTln(K_p)$$ and this is how the pressures/concentrations come into the equilibrium constant.
The transition state is reached only when the reactants have sufficient energy to do so. By the Boltzmann distribution the number of energetic molecules rapidly diminishes as the energy is increased. Thus most collisions do not result in reaction, unless the activation barrier is very small compared to average (thermal ) energy. The reverse reaction has a larger barrier to surmount if the reaction is exothermic, thus fewer collisions result in reaction back to products than the other way round.
The transition state is at the top of the barrier between reactants and products, in solution it may be crossed and recrossed many times before reaction is complete, it all depends upon how quickly the energy the 'molecule' has at the transition state can be removed by collisions or by energy being redistributed internally into the products as vibrations and rotations or as kinetic energy of separation.
The activation energy barrier exists because the product will have different arrangement of bonds and atoms from the reactant and so different average internuclear separation in the reaction coordinate. The activation energy can be imagined as the point where two potential energy parabolas cross, one being slightly displaced in position and energy vs. the other. This is what the normal reaction profile drawn in text books and elsewhere represents. It is usually shown as one line(the bottom of the first parabola), a curved line to a maximum (crossing point, transition state) then onto another lower line to represent product energy at the minima of the second parabola.