Close substitutes vs. perfect substitutes
For me, pluots and grapes are close substitutes, though not perfect substitutes.
Let's say a house visitor brings a big bag of pluots as a gift. That distorts my optimum pluot-grape ratio. But I don't eat all the pluots at once, simply to restore the preferred ratio. That would be an unbalanced meal. Instead, I dig into the pluot bag at a higher rate than I would normally eat pluots. Over a longer period of time the proper ratio is restored, pluot by pluot.
The closer pluots and grapes are as substitutes, the longer it takes for equilibrium restoration. But the closeness of substitution is not always the major factor determining the speed of adjustment. It may be just as important, or more important, how hungry I am, how much of a glutton I am, and how much I like pluots. It also matters whether I am fearing a large shock, requiring a major pluot reserve, such as when a bevy of pluot-devouring guests might suddenly visit the house.
In any case, equilibrium restoration eventually comes. Those pluots will clear my refrigerator, sooner or later.
Apply a similar analysis to cash and T-Bills. At low nominal interest rates they are close substitutes but not perfect substitutes. T-Bills are of greater use for clearinghouse collateral, and yield a bit more, but they are not liquid in exactly the same ways. Even if cash and T-Bills are close substitutes, monetary policy will change the banks' mix and so banks eventually will reequilibrate, converting extra cash into AD, one way or another. But as with the pluot adjustment, this won't happen all at once.
Equilibration (and the accompanying AD boost) may be slow, maybe too slow for comfort. But the degree of substitutability is not the only factor determining the speed of this adjustment, just as with the pluots. The degree of bank and business confidence may be the most important factor. Other factors will be the cues from the Fed, the corporate cultures of the banks, possible uncertainty, transactional frictions (the difference in returns is small but perhaps the cost of adjusting is small too), and so on. The degree of substitutability is one factor determining the speed of adjustment but not necessarily a major factor. I am not sure it should be put in the top three most important factors.
The zero bound argument assumes that the degree of substitutability is, at current margins, the major factor which matters. You are being "sold" a model of portfolio adjustment with one dominant factor.
Monetary policy is not working well today, but perhaps bank and business confidence is a bigger reason than the degree of substitutability. From those who promote the relevance of the "zero bound" argument, you will see it implicitly assumed that we have a case of perfect substitutes and that substitutability is the major factor hindering portfolio adjustment, as would lead to an expansion of the broader money supply aggregates.
In many models "close substitutes" and "perfect substitutes" have different very properties. Current discussions of the liquidity trap usually neglect this point.
Addendum: This is also a framework for thinking about ceasing to pay interest on reserves.