ConfigDB multi threading enhancements

[uw4]

this scenario could happen:

• a save request is issued via thread / request 1
• the request has not written the file fully
• a read request is issued via thread / request 2
• we get a read error on the partially written file

all in all this seems very unlikely. If it happens, it can be "fixed" by a simple reload

introducing file locks would be a possibility, but seems risky, since the behaviour is OS dependent

https://docs.oracle.com/javase/8/docs/api/java/nio/channels/FileLock.html