Currently, NEO-STRESS often fails because it is assigned
on the same test node as other test suites that:
- overload the test node;
- are often triggered at the same time.

The original work for [monitor to plugin migration](nexedi/slapos!614) missed
important parts of resilient stack implementation due to lack of SlapOS
Profile Integration test coverage for those cases.

/reviewed-on nexedi/slapos!659

When the image is a block device, the guest must also run Linux >= 5.0.

BT were moved into bt5/ subdirectory in
nexedi/erp5-doc@3996ae1f

Unfortunately pbs recipe is just simply creating an executable to check own
state, so keep the etc/promise directory in the partition for now, as it is
supported by monitoring checks.

This reverts commit dcd2f8d6.

It has to be redone with proper test coverage.

This reverts commit 04ea0535.

It has to be redone with proper test coverage.

Since the introduction of shared parts, there is no guarantee about
the location of sqlite3 binary.

Calling makeModuleSetUpAndTestCaseClass results already with preparing some
environment and also test classes will run the promises, so they would never
execute test_kvm_sanity_check.

So create a method sanityCheck to check the environment and depending of the
result call or not makeModuleSetUpAndTestCaseClass and skipIf test cases
when environment is not sane enough.