[tmpnet] Enable network restart to simplify iteration

[marun]

Why this should be merged

When making changes to avalanchego or VM binaries as part of e2e iteration, it may be useful to restart a running network instead of having to start a new one.

How this works

Adds --restart-network flag for e2e test suite.

How this was tested

CI against regression, manually for usage:

[SynchronizedBeforeSuite]
/home/me/src/avalanchego/master/tests/e2e/e2e_test.go:41
  Loaded a network configured at /home/me/.tmpnet/networks/20240806-105414.807586-avalanchego-e2e
   restarting network
  Started node "NodeID-5drktRkarrPqdHhKKRpcP6iM94ckg5Xhy"
   waiting for node NodeID-5drktRkarrPqdHhKKRpcP6iM94ckg5Xhy to report healthy
  Started node "NodeID-8kXEqbVv9vZMNyUtUTnVcmXJtQvuEPS9X"
   waiting for node NodeID-8kXEqbVv9vZMNyUtUTnVcmXJtQvuEPS9X to report healthy
  network URIs:  [{NodeID:NodeID-5drktRkarrPqdHhKKRpcP6iM94ckg5Xhy URI:http://127.0.0.1:44575} {NodeID:NodeID-8kXEqbVv9vZMNyUtUTnVcmXJtQvuEPS9X URI:http://127.0.0.1:42131}]
  test data server URI:  http://127.0.0.1:40503
[SynchronizedBeforeSuite] PASSED [12.919 seconds]