feat(llmobs): submit span events for chains from langchain

[sabrenner]

Summary

Follow-up to #8695

This PR adds support for submitting LLMObs span events from the LangChain integration automatically if LLMObs is enabled, and if the LLMObs span is sampled. This is accomplished by:

• Setting the span type SpanTypes.LLM on langchain chain APM spans so they are properly processed by the trace processor service in the LLMObs service
• Tagging each chain span with additional _ml_obs.* tags, which get popped from the trace when submitting the data they represent to LLMObs intake through the LLMObs writer

For Reviewers

Most of the LOC are test changes to account for the change in test structure when dealing with variable amounts of expected calls to the LLMObs writer. Additionally, many of the files changed are snapshot files, to account for the change in the span type. Otherwise, the important files changed are:

1. ddtrace/contrib/langchain/patch.py
2. ddtrace/llmobs/_integrations/langchain.py

Additionally, no release notes/changelog, as this is an internal change for submitting span events to LLMObs intake.

Checklist

• Change(s) are motivated and described in the PR description
• Testing strategy is described if automated tests are not included in the PR
• Risks are described (performance impact, potential for breakage, maintainability)
• Change is maintainable (easy to change, telemetry, documentation)
• Library release note guidelines are followed or label changelog/no-changelog is set
• Documentation is included (in-code, generated user docs, public corp docs)
• Backport labels are set (if applicable)
• If this PR changes the public interface, I've notified @DataDog/apm-tees.

Reviewer Checklist

• Title is accurate
• All changes are related to the pull request's stated goal
• Description motivates each change
• Avoids breaking API changes
• Testing strategy adequately addresses listed risks
• Change is maintainable (easy to change, telemetry, documentation)
• Release note makes sense to a user of the library
• 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

[datadog-dd-trace-py-rkomorn]

Datadog Report

Branch report: sabrenner/langchain-span-events-chains
Commit report: 54f0661
Test service: dd-trace-py

✅ 0 Failed, 2852 Passed, 288 Skipped, 1h 18m 42.98s Total duration (6m 11.54s time saved)

[pr-commenter]

Benchmarks

Benchmark execution time: 2024-04-17 19:55:51

Comparing candidate commit 2af0e25 in PR branch sabrenner/langchain-span-events-chains with baseline commit b1dab58 in branch main.

Found 3 performance improvements and 0 performance regressions! Performance is the same for 198 metrics, 9 unstable metrics.

scenario:flasksimple-appsec-get

• 🟩 execution_time [-247.557µs; -183.475µs] or [-3.806%; -2.821%]

scenario:httppropagationextract-all_styles_all_headers

• 🟩 max_rss_usage [-755.572KB; -438.412KB] or [-3.473%; -2.015%]

scenario:httppropagationextract-medium_header_no_matches

• 🟩 max_rss_usage [-759.079KB; -557.785KB] or [-3.462%; -2.544%]

[Yun-Kim]

Awesome work! Other than a couple of style points, I had a couple suggestions on how we should tag chain inputs and outputs.

[sabrenner]

Since CI is a bit funky with skipping tests it shouldn't, here are the results from a local run of the test suite (with Python 3.10)

test_langchain:

test_langchain_community:

where, in each case, the majority of the skipped count is the other test suite that was skipped

[Yun-Kim]

Awesome work!