feat userver: ppa setup

[segoon]

---

Note: by creating a PR or an issue you automatically agree to the CLA. See CONTRIBUTING.md. Feel free to remove this note, the agreement holds.

[segoon]

This failure (https://github.com/userver-framework/userver/actions/runs/34104753032/job/101687096955?pr=1342) is unrelated to PR #1342.

The job fails before userver itself is configured or compiled:

• Conan is building the external dependency mongo-c-driver/1.30.6.

• Its TRY_COMPILE passes a malformed combined argument:
  "-Werror -DCMAKE_CXX_LINK_EXECUTABLE='echo not linking now...'"

• CMake 4.4.3 interprets that as an unknown warning category and aborts. See /home/segoon/Downloads/job-logs.txt:267355.

• That problematic code comes directly from mongo-c-driver 1.30.6.

• The new GitHub runner image ships CMake 4.4.3, confirming this is toolchain/environment drift. Runner image specification.

PR #1342 changes neither conanfile.py:1 nor .github/workflows/ci-conan.yml:1. The job also uses unpinned pip install conan and conan create --update, making it
sensitive to external updates.

Most appropriate resolution: pin CMake below 4.4 for this job, or update/patch the Conan mongo-c-driver recipe. A simple rerun on the same runner image may fail
identically.

[segoon]

This (https://github.com/userver-framework/userver/actions/runs/34104752928/job/101687096275?pr=1342) also appears unrelated to PR #1342, but it is a genuine macOS subprocess-test flap.

Key detail: status.IsExited() == false means the spawned child was terminated by a signal. It does not mean the FD-leak check failed:

• An inherited logger FD makes the child call std::exit(1), which would still produce IsExited() == true.
• Here the child crashed or was killed before returning normally.
• The assertion at core/src/engine/subprocess/process_starter_test.cpp:233 unfortunately doesn’t report status.GetTermSignal(), so this excerpt cannot identify
  the precise signal.

Evidence against PR causality:

• PR feat userver: ppa setup #1342 changes neither the subprocess implementation/test nor .github/workflows/macos.yml:1.

• Its core/CMakeLists.txt changes only installation/config-export declarations, not this test’s build or runtime behavior.

• The base commit, 8675963, is itself titled cc core: fix MacOS related flaps and modifies this exact test’s macOS executable-path handling. That strongly
  suggests this is a residual base-branch macOS flap, or a regression in that recent fix.

• The failure occurred in the separate macos-latest (brew) job, not the Conan job.