Refactor access with proper mutex guard

[influx6]

Currently we have in geth/node/manager.go (see https://github.com/status-im/status-go/blob/develop/geth/node/manager.go#L35-L45), where we have both intention in code to guard synchronous access to most fields through the embedded mutex.

Problem

But we currently have in https://github.com/status-im/status-go/blob/develop/geth/node/manager.go#L69-L71 and in many other areas within code where access to NodeManager.node is initially accessed without mutex guard to ensure synch access.

Issues

We could later face issues with race conditions on access to the NodeManager.node field and other fields that maybe concurrently reacehed directly or indirectly.

Fix

We need to refactor this package to ensure proper synchronouse access to the NodeManager.nodeand other fields as intended, else re-arrange NodeManager struct fields to ensure quick intention has to fields which mutex guards.

[JekaMas]

Actually we already got a lot of dara races in node/manager.go, node/node.go and node/utils.go.
I think we should turn on a data race detector in ci.

[tiabc]

@JekaMas #50

[JekaMas]

I done most of 403 to complite this task:

• NodeManager is simplified
• Some e2e tests are added
• There are no race conditions when running unit tests with -race.
• NodeManager is thread-safe

[mandrigin]

@divan @adambabik closing in favour of status-im/swarms#63