p2p/simulations: fix intermittent data races related to Node.Up field

[acud]

We should introduce finer granularity locking on individual Nodes and introduce getters&setters to mitigate possible data races.

Node.Up is not properly encapsulated; sometimes it's accessed under the protection of Network.lock sometimes it's just read (e.g., TestStartStopNode) directly. => data race

[frncmx]

Node.Up still causes some data-races. Those are harder to reproduce than in #1110.
=> Let's make the encapsulation.

[acud]

@frncmx wasn't this already done?

[frncmx]

Nope. I did not touch the PR which aims to resolve this for a while. That is ethereum/go-ethereum#18976

[acud]

that PR is fully approved. where's the blocker?

[frncmx]

As there is a suspected deadlock somewhere. But that might have been fixed by p2p/simulations: fix a deadlock calling getRandomNode with lock

I'm just about to pick this up again to make it to the finish line. Thanks for pinging.