Verify the process snapshot-to-telemetry path from a clean clone #6

Review target

Independently reproduce the checked-in linux-process-observe snapshot -> diff -> telemetry adapter workflow from a clean clone. This is a verification task after the temporal-cohort checks added in #4, not a request for a new feature.

Relevant paths on the default branch:

• projects/linux-process-observe/README.md
• projects/linux-process-observe/src/linux_process_observe/cli.py
• projects/linux-process-observe/src/linux_process_observe/diff.py
• projects/linux-process-observe/src/linux_process_observe/adapters.py
• projects/linux-process-observe/tests/test_cli.py
• projects/linux-process-observe/tests/test_adapter.py
• projects/linux-process-observe/tests/golden/diff/telemetry_events.jsonl

Reproduction

1. Start from a clean clone of the current default branch and record the commit SHA, OS, Python version, and install command.
2. Follow the README workflow with the checked-in baseline and changed fixtures to create both snapshots, the diff, and telemetry_events.jsonl.
3. Compare the generated artifacts with the checked-in golden evidence and run the focused CLI/adapter tests.

The documented happy path should produce stacknil.system-evidence.v1 artifacts and seven deterministic telemetry events. Please report any byte-level or semantic difference rather than updating a golden file.

Evidence to post

• Exact commands and exit codes.
• A concise artifact comparison, including event count and any differing relative paths or fields.
• Focused test results.
• One classification: reproduced as documented, documentation ambiguity, environment-specific limitation, or probable defect.

Change threshold

Post the reproduction evidence before opening a PR. A PR is useful only if the evidence identifies a concrete mismatch; keep any correction to the smallest affected contract, test, or instruction.

Boundaries

• Use only checked-in fixtures and disposable output directories.
• Do not inspect the live host or capture real process/socket data.
• Do not post absolute machine paths, usernames, credentials, or full environment dumps.
• Preserve the same-time cohort checks, strict before/after ordering, deterministic ordering, and existing telemetry schema.