Backport of adc7896 from #14802 to
3.16.

Co-authored-by: Emmett Butler <723615+emmettbutler@users.noreply.github.com>

Backport 7cd4d9c from #14794 to 3.16.

## Description

Currently, in Test Impact Analysis code coverage, we use
`is_user_code()` to decide whether to instrument a module. This function
was conceived for use in non-CI contexts where user code may be
installed in site-packages like external dependencies. This is not
necessary in Test Optimization, where we only want to cover code inside
the current repo (TIA only works with filenames that are relative to the
repo root, and we already set the repo root as the include path of
`ModuleCodeCollector`). Removing this check also resolves a bug where
test packages are incorrectly detected as third-party if a third-party
package contains files under `test/*` or `tests/*` that get installed at
site-packages level (`sumy` and `sendgrid` do this, for instance).

## Testing

Unit tests + manual testing in dogweb.

## Risks

Potentially covering more code than expected. Should not be an issue
because we already exclude site-packages and similar
[here](https://github.com/DataDog/dd-trace-py/blob/vitor-de-araujo%2FSDTEST-2747%2Fcoverage-third-party/ddtrace/internal/coverage/code.py#L53).

## Additional Notes

None.

- fix version computation for backports on 3.16
- update ST version to fix a problem related to changes in the agent.

…shed loading [backport 3.16] (#14840) (#14906)

…port 3.16] (#14886)

Backport 8235d03 from #14876 to 3.16.

## Description

Reverting usage of json helper function from libddwaf due to some
possibly related memory corruption errors.

This is a (only partial) revert of
#14215

Also:
- improve a test file to ensure the waf is creating a proper handle from
this file.

Co-authored-by: Christophe Papazian <114495376+christophe-papazian@users.noreply.github.com>

…16] (#14903)

Backport f11c438 from #14837 to 3.16.

When submitting a job, `runtime_env` could be none in `kwargs`, leading
to an `AttributeError`, see this
[link](https://dd.datad0g.com/llm/distributed-ai/jobs?query=%40component%3Aray%20parent_id%3A0%20job-113464b6-33d0-47&agg_m=count&agg_m_source=base&agg_t=count&colorLegendSort=time&fromUser=false&spanId=5240087393744960581&traceId=68e818d800000000e5341ed29f2823d9&start=1759437349664&end=1760042149664&paused=false).

This PR fixes this issue. It also updates two snapshots that were
missing a span. The tests were xfailing so this was not caught in CI.

Co-authored-by: Louis Tricot <75956635+dubloom@users.noreply.github.com>

…rt 3.16] (#14904)

Backport 109a976 from #14818 to 3.16.

Long running spans were appearing with the operation name as the span
name making the trace hard to interpret until the job finished.

This PR adds the resource name to partial flush of long running spans

Co-authored-by: Louis Tricot <75956635+dubloom@users.noreply.github.com>

…6] (#14902)

Backport a75ed3e from #14871 to 3.16.

If a job was failing during job_submission, the long_running_job span
was not finished causing the span to not leave driveline and creating
inconsistent behavior on the UI.

Co-authored-by: Louis Tricot <75956635+dubloom@users.noreply.github.com>

…14839)

Backport b6d846e from #14834 to 3.16.

## Description

These aren't needed or used anymore, so we can stop providing them.

## Testing

<!-- Describe your testing strategy or note what tests are included -->

## Risks

<!-- Note any risks associated with this change, or "None" if no risks
-->

## Additional Notes

<!-- Any other information that would be helpful for reviewers -->

Co-authored-by: Brett Langdon <brett.langdon@datadoghq.com>

….16] (#14900)

Backport 75a8246 from #14875 to 3.16.

We've tracked down some issues related to broken span hierarchy down to
the way that the gRPC instrumentation deals with entering spans, which
causes the grpc span to leak outside it's call when using the .future()
interface. This PR changes the instrumentation to explicitly only enter
the span when calling the continuations instead of just entering
immediately.

## Description

<!-- Provide an overview of the change and motivation for the change -->

## Testing

I've added tests, but haven't run them for a while and working on
getting it running locally. Will update if I run into any issues

## Risks

<!-- Note any risks associated with this change, or "None" if no risks
-->

## Additional Notes

Co-authored-by: Sam Schlegel <sam.s@discordapp.com>

Signed-off-by: Sam Schlegel <sam.s@discordapp.com>
Co-authored-by: Brett Langdon <brett.langdon@datadoghq.com>
Co-authored-by: Sam Schlegel <sam.s@discordapp.com>
Co-authored-by: Christophe Papazian <114495376+christophe-papazian@users.noreply.github.com>

Backport 492135a from #14955 to 3.16.

## Description

<!-- Provide an overview of the change and motivation for the change -->

We have a bug report that there are recursion errors with wrapt 2.x on
our psycopg integration. I'd still like to study if we can recreate the
error in a test but don't want this investigation to block this known
issue.

## Testing

<!-- Describe your testing strategy or note what tests are included -->

## Risks

<!-- Note any risks associated with this change, or "None" if no risks
-->

## Additional Notes

<!-- Any other information that would be helpful for reviewers -->

Signed-off-by: Juanjo Alvarez <juanjo.alvarezmartinez@datadoghq.com>
Co-authored-by: wantsui <wan.tsui@datadoghq.com>
Co-authored-by: Juanjo Alvarez <juanjo.alvarezmartinez@datadoghq.com>
Co-authored-by: Brett Langdon <brett.langdon@datadoghq.com>

Backport e1b6efa from #14963 to 3.16.

## Description

theacodes/cmarkgfm#84

`cmarkgfm` released a new version which no longer properly installs.
This explicitly installs a known working version.

## Testing

<!-- Describe your testing strategy or note what tests are included -->

## Risks

<!-- Note any risks associated with this change, or "None" if no risks
-->

## Additional Notes

<!-- Any other information that would be helpful for reviewers -->

Co-authored-by: Brett Langdon <brett.langdon@datadoghq.com>

Backport 0b0f0d3 from #14971 to 3.16.

## Description

ddapm-test-agent installs vcrpy which installs "wrapt" with no
constraints. This meant we were still install wrapt 2.0 in our hatch
lint environment causing us to type against wrapt 2 even though we
updated pyproject.toml to be "wrapt>=1,<2".

## Testing

<!-- Describe your testing strategy or note what tests are included -->

## Risks

<!-- Note any risks associated with this change, or "None" if no risks
-->

## Additional Notes

<!-- Any other information that would be helpful for reviewers -->

Co-authored-by: Brett Langdon <brett.langdon@datadoghq.com>