remove support for running python system tests with legacy test runner

Currently, python system tests must be designed in a way that is compatible with two different modes of operation - the legacy runner and the pytest runner. Their design philosophies are sufficiently different to induce a lot of friction points and various compatibility issues.

In order for us to be able efficiently write and extend the capabilities of the pytest runner as well as take advantage of all its possibilities, it's not feasible to keep the compatibility layer which enables the python tests to be executed with the legacy runner.

Instead, python tests should be exclusively executed by the pytest runner, providing a predictable environment and behavior.

Legacy runner can still be used to run shell system tests.

Related #3810