I'm using Karate, and I need to prove than an API is (im)properly working when receiving certain concurrent requests, so I'd need to perform the following sequential steps:

• SETUP: Perform some request to create some shared resource in the backend (a Scenario? requests in the Background section?)
• TEST:
  • Perform several requests in parallel somehow modifying that shared resource (a group of Scenarios in a feature, or maybe a data driven Scenario Outline with Examples run in parallel?)
  • Check that the shared resource has the expected data after all the concurrent requests have finished (another Scenario?)
• (possibly)TEAR DOWN: Remove the shared resource and any other resources created by the concurrent requests

Can you do this with Karate? How? If not, what other tool would you recommend to do this?