Introduces a build and test GH workflow

[ignasi35]

References #1324 (built on top of #1326 so the PR runs the new workflow)

This is part of the migrations:

1. from Travis to GH Actions (Migrate from travis to gitHub Actions #1324)
   1. migrate compile and test
   2. migrate release jobs
2. from Bintray to Sonatype (Migrate to Sonatype from bintray #1323)

This first PR is down-scoped to migrating the build and test jobs (1.1).

Once these compile and test jobs (1.1) in GH Actions are stable, I would migrate the release jobs from Travis/Bintray to GHActions/Sonatype in a single PR. That is, I would do 1.2. and 2. at once.

[ignasi35]

In the build for commit 37efe93 there seems to be a consistent failure in tests for NonBalancingIntegrationSpecAkkaHttp:

[INFO] [04/14/2021 13:02:30.167] [pool-8-thread-4-ScalaTest-running-NonBalancingIntegrationSpecAkkaHttp] [akka.actor.ActorSystemImpl(NonBalancingIntegrationSpec)] Starting test: re-discover endpoints on failure
[INFO] [04/14/2021 13:02:30.178] [pool-8-thread-4-ScalaTest-running-NonBalancingIntegrationSpecAkkaHttp] [akka.actor.ActorSystemImpl(NonBalancingIntegrationSpec)] Sending requests to server 1
[INFO] [04/14/2021 13:02:35.201] [pool-8-thread-4-ScalaTest-running-NonBalancingIntegrationSpecAkkaHttp] [akka.actor.ActorSystemImpl(NonBalancingIntegrationSpec)] Sending requests to server 2
[info] NonBalancingIntegrationSpecAkkaHttp:
[info] Using pick-first (non load balanced clients)
[info] - should send requests to a single endpoint (424 milliseconds)
[info] - should send requests to a single endpoint that is restarted in the middle (335 milliseconds)
[info] - should re-discover endpoints on failure (5 seconds, 51 milliseconds)
[info] - should select the right endpoint among invalid ones *** FAILED *** (5 seconds, 25 milliseconds)
[info]   A timeout occurred waiting for a future to complete. Waited 5000000000 nanoseconds. (NonBalancingIntegrationSpec.scala:145)
[info]   org.scalatest.concurrent.ScalaFutures$$anon$1$$anon$2:
[info]   at org.scalatest.concurrent.ScalaFutures$$anon$1.futureValueImpl(ScalaFutures.scala:339)
[info]   at org.scalatest.concurrent.Futures$FutureConcept.futureValue(Futures.scala:476)
[info]   at org.scalatest.concurrent.Futures$FutureConcept.futureValue$(Futures.scala:475)
[info]   at org.scalatest.concurrent.ScalaFutures$$anon$1.futureValue(ScalaFutures.scala:281)
[info]   at akka.grpc.scaladsl.NonBalancingIntegrationSpec.$anonfun$new$11(NonBalancingIntegrationSpec.scala:145)
[info]   at akka.grpc.scaladsl.NonBalancingIntegrationSpec.$anonfun$new$11$adapted(NonBalancingIntegrationSpec.scala:144)
[info]   at scala.collection.immutable.Range.foreach(Range.scala:190)
[info]   at akka.grpc.scaladsl.NonBalancingIntegrationSpec.$anonfun$new$10(NonBalancingIntegrationSpec.scala:144)
[info]   at org.scalatest.OutcomeOf.outcomeOf(OutcomeOf.scala:85)
[info]   at org.scalatest.OutcomeOf.outcomeOf$(OutcomeOf.scala:83)
[info]   at org.scalatest.OutcomeOf$.outcomeOf(OutcomeOf.scala:104)
[info]   at org.scalatest.Transformer.apply(Transformer.scala:22)
[info]   at org.scalatest.Transformer.apply(Transformer.scala:20)

It has failed in all 4 build-test jobs (scalaVersion-jdkVersion matrix).

---

The following build (commit b9c2c25) passed 3 of the 4 build-test jobs.

I think NonBalancingIntegrationSpecAkkaHttp is a flaky test.

[ignasi35]

PS: Using GH Actions is a lot faster than using Travis. By the time all jobs for commit b9c2c25 had completed in GHActions, neither of the travis builds had yet started.

[ignasi35]

Let's see if that makes the tests a bit less flaky on GHActions

[ignasi35]

I think the main blocker for this to move on is the flakiness of NonBalancingIntegrationSpecAkkaHttp.

[raboof]

(took the liberty of rebasing on top of #1331 to see how that affects this PR)

[raboof]

Ironic that on this PR introducing GH Actions, the travis tests are red (specifically NonBalancingIntegrationSpec.scala:59 - perhaps 3 seconds is a bit tight still, will widen)

[raboof]

Do we still need this?

[ignasi35]

Not anymore, no.