chore(llmobs): tli add apm trace

[jsimpher]

https://docs.google.com/document/d/1ZkKRU0JfljPiwLH1mXaAYRpFaEMFpRGKc5axOjo8mjo/edit?tab=t.0
The goal here is to separate the concept of an llmobs trace id from that of an apm trace id, which will have value in trace level insights, as well as resolving some ui issues around multiple llmobs traces sharing a single trace id.
This is accomplished by adding an llmobs trace id tag to the span, passing it down from parent to child, and ultimately using it instead of trace_id to fill the llmobs_span['trace_id'] slot. the original apm trace id now to be stored in ._dd.apm_trace_id.
QA kind of important here, relatively high blast radius, could ruin trace grouping/heirarchy

Checklist

• PR author has checked that all the criteria below are met
• The PR description includes an overview of the change
• The PR description articulates the motivation for the change
• The change includes tests OR the PR description describes a testing strategy
• The PR description notes risks associated with the change, if any
• Newly-added code is easy to change
• The change follows the library release note guidelines
• The change includes or references documentation updates if necessary
• Backport labels are set (if applicable)

Reviewer Checklist

• Reviewer has checked that all the criteria below are met
• Title is accurate
• All changes are related to the pull request's stated goal
• Avoids breaking API changes
• Testing strategy adequately addresses listed risks
• Newly-added code is easy to change
• Release note makes sense to a user of the library
• If necessary, author has acknowledged and discussed the performance implications of this PR as reported in the benchmarks PR comment
• Backport labels are set in a manner that is consistent with the release branch maintenance policy

[github-actions]

CODEOWNERS have been resolved as:

ddtrace/llmobs/_constants.py                                            @DataDog/ml-observability
ddtrace/llmobs/_llmobs.py                                               @DataDog/ml-observability
tests/llmobs/_utils.py                                                  @DataDog/ml-observability
tests/llmobs/test_llmobs.py                                             @DataDog/ml-observability
tests/llmobs/test_llmobs_service.py                                     @DataDog/ml-observability
tests/llmobs/test_propagation.py                                        @DataDog/ml-observability

[github-actions]

Bootstrap import analysis

Comparison of import times between this PR and base.

Summary

The average import time from this PR is: 245 ± 4 ms.

The average import time from base is: 242 ± 2 ms.

The import time difference between this PR and base is: 3.3 ± 0.1 ms.

Import time breakdown

The following import paths have grown:

ddtrace.auto 0.698 ms (0.28%)

ddtrace 0.478 ms (0.19%)

ddtrace._logger 0.389 ms (0.16%)

ddtrace.internal.telemetry 0.389 ms (0.16%)

ddtrace.internal.telemetry.writer 0.195 ms (0.08%)

http.client 0.169 ms (0.07%)

email.parser 0.096 ms (0.04%)

email.feedparser 0.096 ms (0.04%)

email._policybase 0.096 ms (0.04%)

email.utils 0.096 ms (0.04%)

datetime 0.096 ms (0.04%)

_datetime 0.096 ms (0.04%)

ssl 0.073 ms (0.03%)

ddtrace.internal.runtime 0.026 ms (0.01%)

uuid 0.026 ms (0.01%)

ddtrace.settings._agent 0.194 ms (0.08%)

ddtrace.settings 0.194 ms (0.08%)

ddtrace.settings.http 0.168 ms (0.07%)

ddtrace.internal.utils.http 0.114 ms (0.05%)

email.encoders 0.114 ms (0.05%)

base64 0.114 ms (0.05%)

ddtrace.internal.utils.cache 0.054 ms (0.02%)

inspect 0.054 ms (0.02%)

ddtrace.settings.integration 0.026 ms (0.01%)

ddtrace.vendor.debtcollector 0.026 ms (0.01%)

ddtrace.vendor.debtcollector.moves 0.026 ms (0.01%)

wrapt 0.026 ms (0.01%)

wrapt.__wrapt__ 0.026 ms (0.01%)

wrapt.wrappers 0.026 ms (0.01%)

ddtrace.trace 0.050 ms (0.02%)

ddtrace._monkey 0.039 ms (0.02%)

ddtrace.appsec._listeners 0.039 ms (0.02%)

ddtrace.internal.core 0.039 ms (0.02%)

ddtrace.internal.core.event_hub 0.039 ms (0.02%)

ddtrace.bootstrap.sitecustomize 0.220 ms (0.09%)

ddtrace.bootstrap.preload 0.220 ms (0.09%)

ddtrace.internal.products 0.163 ms (0.07%)

importlib.metadata 0.163 ms (0.07%)

importlib.abc 0.107 ms (0.04%)

importlib.resources 0.107 ms (0.04%)

importlib.resources._common 0.107 ms (0.04%)

ddtrace.internal.remoteconfig._connectors 0.029 ms (0.01%)

ctypes 0.029 ms (0.01%)

_ctypes 0.029 ms (0.01%)

ddtrace.internal.remoteconfig.worker 0.028 ms (0.01%)

The following import paths have shrunk:

ddtrace.auto 1.180 ms (0.48%)

ddtrace 0.643 ms (0.26%)

ddtrace.bootstrap.sitecustomize 0.537 ms (0.22%)

ddtrace.bootstrap.preload 0.537 ms (0.22%)

ddtrace.internal.remoteconfig.client 0.537 ms (0.22%)

[pr-commenter]

Benchmarks

Benchmark execution time: 2025-05-30 21:51:06

Comparing candidate commit 8acddb0 in PR branch jsimpher/tli-add-apm-trace-id with baseline commit 7477e5f in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 503 metrics, 5 unstable metrics.

[jsimpher]

first time really playing with sdk so would also appreciate guidance on best practices/any additional tests. nervous ill break somethings so any qa approaches i wouldn't think of please let me know

[yshapiro-57]

I have taken a look at this PR and left comments, feel free to address only those that make sense to you, @jsimpher.

I am doing a low-level line-by-line review here, probably similar to what Cursor would have done. I don't have a clear picture in my head of how different components of this code fit together; for example, if there was a file that was supposed to be changed but was not, I am not going to catch that in my review. So if you are able to please take a look at well, @sabrenner, I will really appreciate it.

[yshapiro-57]

Let me re-approve this now to indicate that my comments have all been addressed. You may want to take a look at the CI errors while we wait for Sam to give this a look as well.

[sabrenner]

looks great!! just doing some manual QAing and might add some other cleanup comments after but i think the approach is solid 😎

[sabrenner]

finishing up QAing but did a quick style pass again, just a couple thoughts