Possibly flaky event-driven I/O DTLS test

[Patater]

We've seen a failure on Travis CI on a PR that only changed comments. We aren't sure how often this error happens, and it might be a flaky test if it sometimes passes.

The test is the "Event-driven I/O, DTLS: session-id resume" test.

• The PR was Iotcrypt 237 hkdf doc #1920
• log.txt from Travis CI

Expected behavior:

• Test should either consistently fail or consistently pass
• Whatever caused this failure should be fixed if a legitimate issue

[ciarmcom]

ARM Internal Ref: IOTSSL-2518