Fix SACK generation when dealing with SYN segments carrying user data

[tuexen]

Summary

When handling a SYN segment with user data in the SYN-SENT state, a DSACK is generated, which is incorrect. This patch fixes this.

Test Plan

Run the following tests from the TCP state event test suite:

• state-event-engine/rcv-syn-fin-syn-sent/rcv-syn-fin-with-data-syn-sent-dropping-disabled-ipv4.pkt
• state-event-engine/rcv-syn-fin-syn-sent/rcv-syn-fin-with-data-syn-sent-dropping-disabled-ipv6.pkt
• state-event-engine/rcv-syn-syn-sent/rcv-syn-with-data-syn-sent-ipv4.pkt
• state-event-engine/rcv-syn-syn-sent/rcv-syn-with-data-syn-sent-ipv6.pkt

[rscheff]

Thanks for updating the packetdrill scripts to match recent improvements too.

[rscheff]

Gave this a 2nd thought, but still think that the check if the session state has transitioned to established overall is the correct approach.

Or would there be any conceiveable paths, where either a DSACK or SACK could be triggerend from a data-SYN?

(Only a DSACK on the receipt of a duplicate data-SYN comes to mind, or when the data-SYN chances in size between the first observed / 2nd receipt. But in these cases, the session would be established too...)