Immediate disclaimer: I took the drawings from a google search and added stuff.
Here you go. A quick couple of sketches. A typical R/R wiring for our bikes looks like this:
The three coils "fire off" at different times and are designed to not fully be in sync with any other. To get a rectified curve, the six diodes are placed where they are to limit the direction of current passed by each coil so that you don't see odd voltages at the battery. A properly functioning R/R should look something like this:
**For simplicity I ignored the third phase**
You can see that when one phase is passing positive current, the other is pulling negative but through the negative end of the battery creating an overall voltage drop at the terminals which facilitates charge. The third is somewhere in between but again, this is an overly simplified sketch but the idea is the same.
When one of the diodes short, you get the following behavior:
For this sketch I shorted out the top right most diode. The coil associated with the green path now draws current from the path of least resistance which happens to be the other phases whose diodes are still working fine. The current in the red path is amplified by the leaking green phase. Because the voltage at the coils are much higher than those at the battery, even through the regulator, the green current tends to keep leaking towards the lower voltage at the positive terminal of the battery for that particular cycle.
And this is the part that I'm not too familiar with. Some regulators only short one phase but others short all to control the output voltage. I suspect the CBR R/R only regulates one phase and if the short happens on the other two unregulated phases, you would see over charging of the battery and a massively fluctuating voltage at the terminals. Or if indeed all three phases are regulated, the increased current through the amplified phase by the leaking phase could just be "overwhelming" the regulator.
What you would see when you measure AC voltage across the terminals of the battery would be the voltage spikes and drops from the leaking current because of a shorted diode.
Having said all of that, cruddy grounds and other connections at the wires anywhere between the alternator coils and the battery terminals could wreak all sorts of havoc also. I suggest carefully tracing the wires along with what I said earlier to find the issue. You've got some gremlins alright.
Also, I just realized I forgot to add green path to the last drawing coming from the negative terminal but that doesn't change what I've said above at all.