Isolate integration test runs by branch

[bart-vmware]

Integration tests for Steeltoe samples run on PRs and on the main/latest branch. Although we recently randomized the routes, they all run in the same CloudFoundry space with fixed names, which means they influence each other.

One solution is to run them in different CF spaces, for example in /Home/steeltoe-samples/branch-name instead of the current /Home/steeltoe-samples.

[TimHess]

I don't think running in parallel is practical for how we're currently running tests since the path filters aren't working and we don't even have capacity to simultaneously run one round of tests

[bart-vmware]

This is about isolation, not parallelism. Samples tend to create a database. If the schema differs per branch, it won't succeed because the existing schema is incompatible. Using different spaces solves that.

Path filters are supposed to address parallelism, that's unrelated to this issue.

[TimHess]

This is about isolation, not parallelism. Samples tend to create a database. If the schema differs per branch, it won't succeed because the existing schema is incompatible. Using different spaces solves that.

Path filters are supposed to address parallelism, that's unrelated to this issue.

Unrelated until we run out of capacity anyway...

I tried a strategy for this in #361, but due to PR shuffling we'll want to re-apply the effort, ensuring to delete spaces after the run (as described in a comment on that PR)