optimize io capture setup/teardown

this is a followup to #401

one perceived issue was that

fd-level capturing takes >10 percent. It's performed for each setup, call and teardown phase of a test and it's known to not be very efficient because it tries too hard to remain compatible to pytest-capturelog. I believe we should aim at integrating the pytest-capturelog plugin and simplify and speed up capturing.

there are various way to work with this

many pure python projects for example don't need fd capture and would benefit from opting for sys-capture per default

we could for example add a ini option to select the capture mode

also different strategies for setting up the fds and the related objects could help in general

Start by reviewing the existing pytest I/O capture implementation and the follow-up context in issue #401. Compare the proposed capture modes and file-descriptor setup strategies, then establish benchmarks and a concrete acceptance scope before attempting the refactor.