Websocket error propagation lacks usability

[btschwertfeger]

When using the Spot websocket client (maybe with the Futures client too), it is not that clear when and why a reconnect is done.

When a reconnect is performed, first, the error is logged via a warning, e.g. keepalive ping timeout; no close frame received or got an exception received 1001 (going away) CloudFlare WebSocket proxy restarting; then sent 1001 (going away) CloudFlare WebSocket proxy restarting (both with traceback). But those are quite useless information when using in an external trading algorithm. Since this is a warning, and reconnect only logs via info level, there is no information about the time of reconnect.

What I propose to improve:

• Make the logging more sensible, meaning that when a reconnect is done, the successful reconnect should be visible on the same level of logging, as the failure that caused it.
• When a reconnect is done or a failure which is to be healed via reconnect is occurred, there should be some mechanism that packages using this functionality can react on that, e.g. if the execution channel reconnects, than we have to ensure that all executions, even those that occurred during the reconnect are handled properly.