fix(dcutr): Fix end to end tests and add legacy behavior flag (default=true)

[MarcoPolo]

Previously this PR sought to migrate away from the legacy behavior. Now it just tests the hole punching behavior using a simulated network. It also adds a flag to control the legacy behavior, but this is set to default to the current behavior.

Most of the code introduced here is related to the new simconn package which lets us create simulated UDP connections for use with the QUIC transport. This lets us set up arbitrary networks to test holepunching. The End to End test now places hosts A and B behind a firewall that requires holepunching for a direct connection.

The plan is to merge this PR to actually test holepunching, then #3171 by coming up with a migration strategy and use these tests to verify the strategy.

[sukunrt]

On the side that in the new code will be the Server, can we add a random wait time after which the server will switch the role to Client and initiate the security negotiation?

[MarcoPolo]

Does it need to be random?

Could the server wait 3 RTT (it has the estimate) for the client and switch after?

[sukunrt]

Unfortunately simple timeouts wont work. This is trickier.

We need to always start as a Server. Then switch roles after a random wait. This is because when old go nodes and new go nodes interact, they may both assume the Client role, in which case, conn establishment will fail as both the sides assume the Client role.

In the new code, if we always assume the role of Server, there will be 3 cases:

old go node vs new go node

1. Client <-> Server:
   For client client to work, we need to assume the role of Server in new code. The peer, old go node, will initiate the handshake.
2. Server <-> Server
   For server server to work, we need to assume the role of Server in new code, switch roles and initiate the handshake.

new go node vs new go node

1. Server <-> Server:
   After a random backoff one side becomes the client and initiates the handshake.

[sukunrt]

Some minor comments. It is fine to merge for me. This is a significant improvement over the current testing of the holepunch package! 🚀

[sukunrt]

Should we make the ConnManager an interface so that we have a simple implementation to use simulated conns.
Admittedly, it'll not test the full holepunching stack. But the current system feels a bit too complex.

I don't have a strong opinion on fixing this right now.

[MarcoPolo]

I don't think I follow, can you explain a bit more?

[MarcoPolo]

Oh, I think I understand. You're suggesting having quicreuse implement some new interface that we could then implement some simpler mock thing that returns the simulated packet conns.

If that's it, I think I'd prefer to have these options in quic resuse for the following reasons:

1. This doesn't change any of the real logic of the quicreuse package.
2. It keeps the simulated environment closer to the real environment.
3. We could test how QUIC and WebTransport interact in quicreuse with holepunching.