Help the client recover from rejected phantom connections due to station load

Due to high load at the station, the client sometimes receives the following error when attempting to connect to negotiated proxies:

[11:32:12] [1-d9b572] failed to dial phantom [scrubbed]: dial tcp [scrubbed]: i/o timeout

Instead of rejected the SOCKS connection, we could have multiple retries. The gotapdance library already has some mechanism for spacing out retries I believe.