chore: enable shuffle testing for most core packages

[felixge]

What does this PR do?

Enable test shuffling for most of our core packages.

Packages that were unable to successfully pass their tests with -shuffle on for 10 times in a row (on my local machine) have been excluded. See #4060 for details.

Additionally the following packages were excluded because they flaked in CI (this may or may not be due to shuffle):

• instrumentation/appsec/dyngo
• instrumentation/httptrace

The plan is to enable shuffling for the excluded packages in the future once their test suite has been fixed. Enabling shuffle testing for contrib packages will also be done separately.

If you see test flakes, and suspect that they are caused by this PR, please try to fix them instead of adding the packages to the shuffle exclusion list.

PROF-12813

Motivation

Some of our tests have come to rely on global state pollution over time. I.e. they need another test to run first in order to pass, or rely on another test not running first in order to pass. This is undesirable as it often causes problems for people debugging test failures (e.g. test passes when go test -run TestXXX but not when running all tests in the pkg). It may also hide implementation bugs.

Reviewer's Checklist

• Changed code has unit tests for its functionality at or near 100% coverage.
• System-Tests covering this feature have been added and enabled with the va.b.c-dev version tag.
• There is a benchmark for any new code, or changes to existing code.
• If this interacts with the agent in a new way, a system test has been added.
• New code is free of linting errors. You can check this by running ./scripts/lint.sh locally.
• Add an appropriate team label so this PR gets put in the right place for the release notes.
• Non-trivial go.mod changes, e.g. adding new modules, are reviewed by @DataDog/dd-trace-go-guild.

Unsure? Have a question? Request a review!

[pr-commenter]

Benchmarks

Benchmark execution time: 2025-10-22 16:11:33

Comparing candidate commit 685c422 in PR branch felix.geisendoerfer/chore-enable-shuffle-testing-for-most-core-packages with baseline commit be655e9 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 24 metrics, 0 unstable metrics.

[nsrip-dd]

I confirmed that the tests are split up appropriately in the CI run: https://github.com/DataDog/dd-trace-go/actions/runs/18722222366/job/53397550605?pr=4061#step:6:237

I also confirmed locally that gotestsum will log the shuffle seed if the tests fail:

% gotestsum -- -v -shuffle=on ./profiler
✖  profiler (21.383s) (-test.shuffle 1761161421624258000)

=== Skipped
=== SKIP: profiler TestDebugCompressionEnv (0.00s)
    compression_test.go:77: Flaky. See #3681

=== Failed
=== FAIL: profiler TestFail (0.00s)
    profiler_test.go:948: fail!

DONE 152 tests, 1 skipped, 1 failure in 21.383s

Thanks for doing this!

[felixge]

/merge

[dd-devflow-routing-codex]

View all feedbacks in Devflow UI.

2025-10-23 04:05:07 UTC ℹ️ Start processing command /merge

---

2025-10-23 04:05:12 UTC ℹ️ MergeQueue: pull request added to the queue

The expected merge time in main is approximately 18m (p90).

---

2025-10-23 04:20:04 UTC ℹ️ MergeQueue: This merge request was merged