This branch is going to fix a lot of issues for quite some people.

Fixes #569, #587, #566, #586 and #280.
This includes the fixes as suggested in #671, #615 and #533. Thx a lot to @qdb, @tobek and @chrishamant
Closes #675, #463, #362, #438 and #724

There will be more tickets that we closed without fixing them that get fixed by this PR.

This is going to do the following:

1. [FIX?] Change the default connection timeout to be one day instead of never.
2. [FIX] There won't be any connection retries after the connection timeout anymore. This was the case earlier because the timeout for the next retry is increasing each retry by a multiplier. If the next retry would be above the connection timeout it is now going to be exactly at the maximum retry connection timeout instead. So the timeout is followed as expected from now on.
3. [??] The default minimum connection retry time is reduced a tiny bit (200 ms instead of 255).
4. [FIX] The queues won't be flushed anymore if node_redis is trying to reconnect. It will only be flushed if the connection is broken or if the user closes the connection. So offline_queue and max_attempts / connection_timeout is now properly working together.
5. [FEATURE] If no connection could be established and max_attempts / connection_timeout has been exceeded node_redis is going into the "connection is broken" mode and emit an error.

This should be ready to merge