Fix for the race condition being created when two threads were simultaneously doing refresh

[Sfurti-yb]

In the Refresh() function initialHosts list is being read by one thread while the other thread might be removing an unreachable host from it, mutating the list.

 internal override bool Refresh()
    {
       
        initialHosts = initialHosts.Distinct().ToList();
        foreach (var host in initialHosts.ToList())
        {
            try
            {
                // Setting up control connection with host and doing refresh 
            }
            catch (Exception)
            {
               // if host connection fails remove it from the initialHosts list
                _connectionLogger.LogDebug("Failed to connect to {host}", host);
                initialHosts.Remove(host);
                if (initialHosts.Count == 0)
                {
                    _connectionLogger.LogError("Failed to create control connection. No suitable host found");
                    throw;
                }

            }

        }
        unreachableHostsIndices.Clear();
        unreachableHosts.Clear();
        return true;
    }

This is creating the following race condition and throwing the error:

System.ArgumentException: Source array was not long enough. Check the source index, length, and the array's lower bounds. (Parameter 'sourceArray')
   at System.Array.CopyImpl(Array sourceArray, Int32 sourceIndex, Array destinationArray, Int32 destinationIndex, Int32 length, Boolean reliable)
   at System.Collections.Generic.List`1.set_Capacity(Int32 value)
   at System.Collections.Generic.List`1.AddRange(IEnumerable`1 collection)
   at YBNpgsql.ClusterAwareDataSource.Refresh()
   at YBNpgsql.NpgsqlConnection.SetupDataSource()

This fix sets up a lock before needsRefresh() is called so that only one thread actually does the refresh, the other threads can move past this step.

Testing

Tested it with a multithreaded application that was long running and read from the database. A shell script was used to periodically down and restart the server nodes.