fixed the failing persistence test by re-ordering the clients list.

The test fails because initialization of all EL clients takes longer
than the initial genesis delay.
Due to this, lighhouse returns an unhealty result for its health check
as it's post-genesis and not following a chain / has no peers.
The kurtosis package waits for a healthy result, which never occurs as
all other BNs & VCs are missing at that point.

The fix proposed in this PR is a workaround as it moves the lighthouse
node further down to ensure it can connect to other nodes initialized
before.

There is an alternative approach to fix this issue:
```
network_params:
  genesis_delay: 300
```
which ensures all clients are properly initialized before the network
starts.