CBL-3666: Connectivity Manager can be registered multiple times.

[bmeike]

Hotfix for CBSE-12499

When a continuous replication goes offline, it registers an observer to watch for changes in connectivity. The code used to allow a single observer to be registered multiple times. If this happened often enough, the system would throw an uncheck exception: TooManyRequestsException, which will cause the thread to fail.

The heart of the fix is line 313, checking to see if the observer is already in the map. The try-catch blocks are just bullet-proofing.

[borrrden]

I'm assuming this is a simple merge of something that already went out

[bmeike]

@borrrden : it is not. I put this fixe in Helium first. It seems to work but, as noted, I do not have a good way to test it very thoroughly. I assumed that we would push the relevant customer to update to 3.0, so I put it in 3.0, next. That has not been released. I just got notification from Dan that we'll need a 2.8 VF. This will produce that VF.