bug fix - V1 canHandle() should cater maxWaitTime

[jayjlu]

The bug was found when we try registering V3 connector with catch all router "*", some requests which should be handled by V1 connectors but randomly being sent to the V3 catch all connector.

The root cause is that there was a bug in the WebsocketFarm (for V1), when it work out whether there are available connector for the current request, it just check the current availability. But given the fact of under traffic load, the V1 connector is possible have some delay on registering back ( as it need to create new connector socket from time to time ).

A test case (src/test/java/com/hsbc/cranker/mucranker/MultiConnectorTest.java) is added to reproduce this consistently.

The fix is to cater the wait time like below

    public boolean canHandle(String target, boolean useCatchAll) {
        final String routeKey = resolveRouteKey(target, useCatchAll);
        if (routeKey == null) return false;
        final Queue<RouterSocket> routerSockets = sockets.get(routeKey);
        if (routerSockets != null && !routerSockets.isEmpty()) return true;
        return routeLastRemovalTimes.containsKey(routeKey)
            && (System.currentTimeMillis() - routeLastRemovalTimes.get(routeKey) < this.maxWaitInMillis);
    }

The PR also including some javadoc fix.