Fix thread leak in DSM specs, clean up FFI thread leak checks

[p-datadog]

What does this PR do?

1. closes DSM Processor after the test to avoid leaking background threads created by the processor
2. adds an exclusion for FFI threads to the thread leak detector
3. removes existing exclusion of detector in a test that triggered FFI threads

Motivation:
Fix https://github.com/DataDog/ruby-guild/issues/274

Change log entry
None

Additional Notes:

How to test the change?

Ran changed files locally to verify no output from thread leak detector

[datadog-datadog-prod-us1]

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage
• Patch Coverage: 100.00%
• Total Coverage: 98.55% (+0.13%)

View detailed report

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: ccf0692 | Docs | Datadog PR Page | Was this helpful? Give us feedback!}

[pr-commenter]

Benchmarks

Benchmark execution time: 2025-11-12 14:50:30

Comparing candidate commit ccf0692 in PR branch dsm-thread-leak with baseline commit bfe50a4 in branch master.

Found 3 performance improvements and 1 performance regressions! Performance is the same for 40 metrics, 2 unstable metrics.

scenario:line instrumentation - targeted

• 🟩 throughput [+9734.307op/s; +10191.881op/s] or [+6.824%; +7.145%]

scenario:line instrumentation - untargeted

• 🟩 throughput [+4948.530op/s; +5064.572op/s] or [+9.200%; +9.416%]

scenario:method instrumentation

• 🟩 throughput [+11941.885op/s; +12492.566op/s] or [+7.268%; +7.603%]

scenario:profiling - Allocations (baseline)

• 🟥 throughput [-315046.008op/s; -300903.759op/s] or [-5.949%; -5.682%]

[ivoanjo]

👍 Left a few notes, but overall it LGTM. Thanks for making this better!