The YouTube video:
https://www.youtube.com/watch?v=qfZ7zyc6ceg

Disclaimer on expected duplicates:

I assert that this is likely same bug as MC-11598 and MC-5864, the first of which I assert to be misunderstood as a duplicate to something it is not and the other not having sufficient information provided and remaining unconfirmed respectively.
Therefore I am posting this to provide more clarification and information.

I assert that this is not the same bug as MC-11193 because, although the bug happens dependent on location, at that given location it happens independently of that location and of the redstone being broken (I don't break any) In fact, at the location, it is completely dependent on the status of the water in/out of the dispenser.

Therefore, I would have to assert that the bug I report as well as MC-11598 are dispenser dependent, redstone independent bugs and cannot be the same as MC-11193 which has no dependence on dispensers but on location and redstone block updates.
It may be argued that the random order of the redstone powering (as shown in MC-11193) occuring to the dispensers caused the behavior. This may explain the extra source blocks being created, but it does not explain why the existing sources, or any, are not retracted by certain dispensers.

Also, this is not a duplicate of MC-4000 which refers to a bug about dispensers with multiple items within. I only am using single water buckets.

Bug: (See video as description is complex) At some locations, when dispensers with buckets inside have water source block in front of them are powered, they do not click audibly and they do not retrieve the water. (Tested with redstone on top of dispensers, wood button signal length and also that with a single tick repeater, in case that effects the signal length at all) Those dispensers, when afterwards are given indirect power from a block being powered, do not click and do not retract the water. However, given once more, they click and retract. (This is all in the video) These dispensers do not have any issues dispensing the water, however, so the bug at the given circuit and location solely depends on the status of the water. Then, given the status of the water, the bug somehow depends on whatever actually sparked the incorrect behavior (i.e. the redstone input) to be fixed with some kind of unknown update (i.e. the indirect power from button powered block).
The extra water sources flowing in the video was a result of two bugged retractions failing around a successful one, creating another source block. I didn't discover that until after the video was created.