Fix sporadic test failure in Elasticsearch53SpringRepositoryTest

[trask]

Hi all!

This is helping to resolve the sporadic test failures in Elasticsearch53SpringRepositoryTest, see open-telemetry/opentelemetry-java-instrumentation#28.

[trask]

🙁 starting to see sporadic failures with this change at:

java.util.concurrent.TimeoutException: Timeout waiting for 2 trace(s). ListWriter.size() == 1
	at datadog.trace.common.writer.ListWriter.waitForTraces(ListWriter.java:43)
	at springdata.Elasticsearch53SpringRepositoryTest.waitForTracesAndSortSpans(Elasticsearch53SpringRepositoryTest.groovy:430)
	at springdata.Elasticsearch53SpringRepositoryTest.test CRUD(Elasticsearch53SpringRepositoryTest.groovy:115)

I will investigate further and post back here.

[randomanderson]

This one seems similar to the sporadic Couchbase test failures that took multiple attempts (#998, #992, #995, #996, #1010, #1021) to get to work.

One change that helped was splitting test CRUD into -> "test save and retrieve", "test save and update", and "test save and delete". (43cbf7a)

One of the other keys found later was that repo.deleteAll() creates a variable number of traces depending on the state of the repo so having it in setup() was impossible. Each test needs to call deleteAll itself and wait for those traces (see most recent version CouchbaseSpringRepositoryTest)

[tylerbenson]

@randomanderson I propose we merge this PR for now and continue working with your suggested fix in a future PR. Objections?

[randomanderson]

@tylerbenson I agree. Sorting the spans is an improvement.