fix(integration): Update PQ integration test expectations

[goatgoose]

Description of changes:

The KMS PQ network integration test recently started failing in CI because KMS started negotiating a different KEM group:

assertion `left == right` failed
  left: Some("SecP256r1Kyber768Draft00")
 right: Some("x25519_kyber-512-r3")

We're currently expecting x25519_kyber-512-r3 to be negotiated, but KMS is now negotiating SecP256r1Kyber768Draft00.

I don't believe this to be indicative of any regression on our side, since this failure wasn't caused by any specific PR. Also, SecP256r1Kyber768Draft00 is actually a more preferred KEM group in the test security policy (PQ-TLS-1-2-2023-10-09):

s2n-tls/tls/s2n_kem_preferences.c

Lines 30 to 32 in c6b41ef

	const struct s2n_kem_group *pq_kem_groups_r3_2023_06[] = {
	&s2n_secp256r1_kyber_768_r3,
	&s2n_x25519_kyber_768_r3,

This PR updates the expectations of this test.

Call-outs:

The purpose of this test is simply to verify that we can successfully perform a PQ handshake. The specific KEM group seems less important. So I updated this test to just check that any kyber KEM group was negotiated. We could instead update this test to check for the new KEM group specifically, if this seems preferable.

Testing:

The PQ integration test should now succeed.

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

[jmayclin]

From the comment above

    // Note: In the future KMS will deprecate kyber_r3 in favor of ML-KEM.
    // At that point this test should be updated with a security policy that
    // supports ML-KEM.

So I'd vote for switching the security policy and then asserting on one of the standardized IANA groups like SecP256r1MLKEM768.

[goatgoose]

I tried this but it only sometimes worked. I think some of the KMS hosts maybe still only support the draft KEM groups. For now I updated the test with a security policy that includes both ML-KEM and kyber groups, and assert that any KEM group was negotiated. I opened an issue to update this test to assert ML-KEM when we're able to: #5086