zephyr-upstream: List of known (aka expected) failure testcases

We have cases of systematic test failures in zephyr-upstream test project (https://qa-reports.linaro.org/lite/zephyr-upstream/). Oftentimes, failure are hardware-specific (happen on one board, but not another).

Ideal solution for these would be of course fixing them upstream, but that may take varying amounts of time and effort, so we need to have a policy and process to deal with them in our testing infrastructure.

An obvious stop-gap measure would be just excluding such tests from LAVA submission. Indeed, we already have some of such tests in the "filtering blacklist". But this is not ideal solution, as we run some arbitrary subset of the upstream testsuite, miss proper metadata of the "known failure", and miss opportunity to see when the issue is fixed.

Instead, SQUAD offers a concept of "known failure" (XFAIL) status for a testcase. Based on quick exchange with @mwasilew, it works like following: it's possible to set a "KnownIssue" field (URL) for a particular testcase and environment (i.e. hardware configuration). If a testcase come in with FAILED status, for which KnownIssue is set, it's instead recorded as XFAIL. This feature is used extensively by LKFT, and based on the above description, would fit out usecase well either. A caveat is that currently, there's no SQUAD UI to manage KnownIssue field, instead only API endpoint is available. LKFT project seems to have an adhoc tool to manage these (looks like functionality isn't available currently in "generic" squad-cli).

Some links:

https://qa-reports.linaro.org/lkft/linux-stable-rc-5.7-oe/build/v5.7.6-265-g97943c6d41ef/testrun/2884169/suite/kselftest/tests/ - example of LKFT usage
https://qa-reports.linaro.org/api/knownissues/ - API endpoint
https://github.com/Linaro/qa-reports-known-issues - existing management tool

Linaro / lite-lava-docker-compose

zephyr-upstream: List of known (aka expected) failure testcases #97