Let's assume your reactive complex (reactants interacting) are optimised, i.e. its gradient equals zero, and your potential energy surface is continuous.
Then any direction is either uphill or constant (otherwise the optimisation hypothesis would not apply). There are three possibilities: (a) going from reactants to intermediate gives a transition state which has higher energy than both reactive complex and intermediate; (b) reactants and intermediate lie on a plateau where energy does not change; or (c) one hypothesis is wrong.
So yes, for barrierless processes.
Completely normal, especially for a bimolecular process forming and adduct like that. However, you shouldn't be applying transition state theory for this kind of process - variational methods (minimising the reaction rate) are more accurate.
These systems can be treated very efficiently using energy grained master equation with RRKM theory, for example software packages look at Mesmer or Paper.