CI testing should catch compile errors in the example code

[roberthbailey]

One way to do this would be to have CI actually run e2e tests that execute the example code (as we do for the simple-udp server). But at minimum we should make sure that the code compiles.

I've recently run across a couple of examples of this:

1. In Updates to the C++ SDK along with the simple example that exercises it. #934 I pushed a change with a typo (book instead of bool), but the build on that hash succeeded. Running make build locally caught the error immediately.

2. I recently tried to grab the rust-simple example from the 0.11.0 release tag to deploy it on a 0.11.0 cluster to see if it used to work (while trying to debug Rust SDK does not wait for connection to be ready #938). Unfortunately, the example code didn't build at the release tag.

[aLekSer]

I prepared a PR with building 5 projects from examples directory, but build now took 30 minutes. As it is hard to calculate hash of all the files in examples folders and understand what was changed from last build, we can call this test randomly for example one out of five pushes, using commit-id for instance. Another option is to remove this new step from cloudbuild or somehow use Prow to make this kind of build testing like nightly builds or by request.

[roberthbailey]

I like the idea of doing this less frequently (since the examples don't change often) to not slow down all PRs.

[github-actions]

'This issue is marked as Stale due to inactivity for more than 30 days. To avoid being marked as 'stale' please add 'awaiting-maintainer' label or add a comment. Thank you for your contributions '

[markmandel]

Added awaiting-maintainer as we should do something here at some point around this.

Maybe it should be conditional on if example code changed?