Fix flaky call.ustack_* tests in CI

[danobi]

The four tests in question:

call.ustack_stack_mode_env_bpftrace
call.ustack_stack_mode_env_override
call.ustack_stack_mode_env_perf
call.ustack_stack_mode_env_raw

They run locally just fine - I couldn't get them to fail even once after many runs.

But in CI, they at least 1 fails pretty frequently. On the order of 1/10 jobs. So once per PR.

I debugged as far down as std::cout having badbit 0 set after a write. It's really confusing why. It cannot be reproduced locally. My current best guess is that somewhere in all the layers of shelling out there's a misbehaving pipe that hangs up too early.

[danobi]

While debugging in ongoing, disable tests in CI for now: #3078

Note "for now" is optimistic. I'm quickly running out of ideas

[danobi]

I still think the real bug is in the runtime test suite somewhere, and related to pipes, but unclear why it only affects ustack tests.