spec test runner

[Andrew Kent (realark)]

add btx spec test runner to the python sdk. For now, the only runner supported is openai.

(side-note: I hooked up anthropic as well, but the specs and the sdk disagree on the span structure. will address this later)

--- usage ---

• run end-to-end and assert against real brainstore spans: nox -s "test_btx_openai(latest)" -- --disable-vcr
  • https://www.braintrust.dev/app/braintrustdata.com/p/python-unit-test/trace?object_type=project_logs&object_id=955087e9-e7ce-4461-9e2c-5b4f5fedde1d&r=c77a32a2-ff83-44ac-9a86-2e5df0a3ee90&s=850a4e23-b450-4445-b809-1b1dfc476680
• run offline with vcr and in-memory span assertions: nox -s "test_btx_openai(1.77.0)" "test_btx_openai(latest)"

[Abhijeet Prasad (AbhiPrasad)]

sorry I just merged in a really big refactor to our testing setup: #300

you'll have to rebase this 😅

[Andrew Kent (realark)]

sorry I just merged in a really big refactor to our testing setup: #300

you'll have to rebase this 😅

haha, no worries