There are multiple ways to get erroneous blood pressure readings. Getting an accurate blood pressure is sometimes tricky.
...is there a considerable store of blood waiting to be released behind the cuff and that this would become greater and at higher pressure, the longer the brachial artery is occluded?
Yes, the pressure in the brachial artery can increase the longer the cuff is kept inflated, but it will be minimal, as at some point, the artery will not accept any more blood flow, and it will be diverted away from the brachial artery. Otherwise, you could rupture an artery, which is not what happens. What occurs if a cuff is kept inflated is ischemia distal to the cuff.
There are two reasons one shouldn't keep the cuff inflated any longer than needed (besides ischemia). Both will cause an increased (unrepresentative) systolic reading. First is the (minimal) build-up of blood, but more importantly (imo) is that it hurts! In fact, the longer the cuff is inflated above the systolic BP, the more it hurts. The higher the cuff is inflated, the more it hurts. The more it hurts, the more adrenaline is released and the higher the systolic pressure gets, until the patient rips off the cuff and leaves, muttering (if not shouting) curses at the sadist taking the BP measurement. At least, that's what I think should happen to careless takers of BP. And the patient will forever after have "white coat hypertension".
Is it the case that the fifth phase only occurs if sufficient pressure is released on the cuff?
Yes. (That should be obvious by the very definition of the phases.)
Does it make sense to go faster than 2-3 mm Hg per second once the first phase has been discovered?
In the hands of someone experienced in taking and interpreting BP measurements, yes. That's because the slower you release air from the cuff, the longer the patient is in some discomfort.* However, when learning, reducing cuff pressure too quickly may cause you to overshoot the moment when the diastolic BP makes itself known (i.e. you may get a falsely low diastolic pressure.) Also, if the patient has a narrow pulse pressure, you may miss the diastolic pressure completely.
To reiterate the converse, the fifth phase will not occur if the cuff pressure is not released significantly, i.e. the blood will continue to leak through on the fifth phase?
Yes. (That should be obvious by the very definition of the phases.) Between the systolic and diastolic pressures, the blood can partially overcome the pressure in the cuff. At first, it's difficult, which is why you hear a distinct sound: it is the moment when, at peak systolic pressure, the blood first passes through the cuff. As pressure is released, it flows strongly, swishing loudly, and softens until there's no sound at all.
*I think the practice of inflating the cuff to 200mmHg before starting to release is ill thought out, and causes completely unnecessary discomfort. Common sense should prevail; if the patient is young and healthy, I started 10-15mmHg above what the average should be. I also let the pressure down a bit faster but slowed it down at around 100-90mmHg. And 99% of the time, that sufficed. If the patient is likely to have HBP, I started higher, but never at 200mmHg. In the ED, hypertensive crises were not uncommon, so SBP measurements started higher.