test(openai): drain completion raw_response streams to fix cassette leak

[Kyle-Verhoog]

What

The same VCR cassette-leak pattern fixed in #18778 for the chat_completion raw-response streaming tests also exists two tests earlier for the plain completion variants:

test_completion_raw_response_stream and test_acompletion_raw_response_stream open completion_streamed.yaml with stream=True via with_raw_response.create(...) but never consume the stream, so the VCR connection is never released. Under pytest-randomly orderings the cassette bleeds into later tests, whose unrelated requests fail to match and raise CannotOverwriteExistingCassetteException — the same failure mode reported as "~63 llmobs rate-limit failures" in the recent dd-trace-py CI incident (mislabeled as external API 429s by automated triage; it's actually this cassette leak).

Fix

Drain the raw stream inside the cassette block (resp.parse() / async for), same approach as #18778.

Testing

Ran the affected tests directly and the full test_openai_v1.py file under the exact pytest-randomly seed (3338286823) that originally triggered the leak in #18778 — all pass, 0 failures.

Claude session: 46d356c1-1ac1-4826-b78b-ea87f4220dda
Resume: claude --resume 46d356c1-1ac1-4826-b78b-ea87f4220dda

[cit-pr-commenter-54b7da]

Circular import analysis

⚠️ Existing circular imports

There are 37 circular imports that already exist on the base branch and have not been changed by this PR.

Show existing cycles (showing 5 of 37 shortest)

ddtrace.internal.datastreams -> ddtrace.internal.datastreams.kafka -> ddtrace.internal.datastreams
ddtrace.internal.core -> ddtrace._trace.span -> ddtrace.internal.core
ddtrace.internal.datastreams -> ddtrace.internal.datastreams.kombu -> ddtrace.internal.datastreams
ddtrace.internal.datastreams -> ddtrace.internal.datastreams.aiokafka -> ddtrace.internal.datastreams
ddtrace.internal.datastreams -> ddtrace.internal.datastreams.botocore -> ddtrace.internal.datastreams

To see all cycles, download the cycles-base.json and cycles-pr.json artifacts from this CI job and run:

uv run --script scripts/import-analysis/cycles.py compare cycles-base.json cycles-pr.json

[cit-pr-commenter-54b7da]

Codeowners resolved as

tests/contrib/openai/test_openai_v1.py                                  @DataDog/ml-observability