Introduce underscore-prefixed keywords for the scenario json data files.
Use such keywords for test project naming first, but pave the way for
other potential use cases. Ways of automatically generating test data
may emerge, beside setting environment data with framework's JAVA_OPTS.
Have the CloneUsingBothProtocols showcase scenario use _PROJECT like so.
Make that scenario enforce its class name as the test project name,
based on that specific underscore-prefixed keyword. The scenario then
automatically names the project it creates, clones and deletes. Make the
project used, named after its originating class, more explicit that way.
Make the CreateProject and DeleteProject scenarios support the PROJECT
name property. That property can optionnally be set through framework's
recent JAVA_OPTS support. However, enforce the aforementioned _PROJECT
automation for CloneUsingBothProtocols's aggregation of these two project
scenarios. To make the whole of that behaviour consistent, introduce the
ProjectSimulation class.
It is still possible for a misconfigured test environment to execute the
CloneUsingBothProtocols scenario with PROJECT name set, say, as in [1].
This leads to CreateProject and DeleteProject using 'project' as the
name of the project they create/delete. CloneUsingBothProtocols, which
aggregates them both as delegate scenarios, then fails to clone its
project automatically named 'CloneUsingBothProtocols'. It is then
necessary to unset [1], or use another aggregate scenario composition.
[1] JAVA_OPTS="-Dcom.google.gerrit.scenarios.project=project"
Change-Id: Ib060ed35c5c19a7f8aff05e8d0b856abde70be2e
b39b794a39