fix(s2n-quic-dc): make TCP shutdown a no-op

[camshaft]

Description of changes:

This change makes TCP shutdown a no-op. This is done for a few reasons:

• We already authenticate stream closures at the record layer so doing it in TCP is redundant
• It avoids an extra syscall

Above all, we can avoid more complex shutdown behavior in the TCP layer:

https://datatracker.ietf.org/doc/html/rfc9293#section-3.3.2

A RST can be sent from any state with a corresponding transition to TIME-WAIT (see [70] for rationale). These transitions are not explicitly shown; otherwise, the diagram would become very difficult to read. Similarly, receipt of a RST from any state results in a transition to LISTEN or CLOSED, though this is also omitted from the diagram for legibility.

Basically, we don't need to go through the 4-way finalization flow but instead jump directly to TIME-WAIT. When combined with linger=0, this can release sockets a lot sooner to the OS which enables busy machines to handle higher TPS (see https://doi.org/10.1109/INFCOM.1999.752180).

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[Mark-Simulacrum]

I think we should also consider the linger=0 change being default, but I think it can happen separately.