feat(tools/e2e): limits harness saturation probe (verdict from an observation window)

[Minipada]

Summary

• Adds tools/e2e/scripts/saturation_probe.py, a pure verdict function for Limits harness: measure where the pipeline saturates, on four axes #323's limits harness: given a chronologically ordered window of ack-latency / unacked-window-depth / disk-buffer / outage-declared observations, decides saturated / not-saturated / insufficient-data and names the binding signal.
• Signal priority matches the PRD exactly: acknowledgement latency crossing its bound and staying there (primary) → unacked-window depth trending upward (secondary) → disk-buffer growth at steady state, i.e. only when no outage is declared in the window (tertiary). Buffer utilisation alone is never a signal — only its growth trend, told apart from a declared outage by the outage_declared flag, matching the PRD's explicit rejection of a level threshold.
• A window shorter than min_observations returns an explicit INSUFFICIENT_DATA verdict, never a default healthy/saturated one.
• Pure function: no containers, no I/O, no network — runs entirely as unit tests.

Closes #377

Test plan

• uv run pytest tools/e2e/test/test_saturation_probe.py -v — 13/13 passing, covering all six acceptance criteria plus signal-priority ordering and the utilisation-vs-trend distinction
• uv run pytest tools/e2e/test/ — full suite (37 tests) still green
• uv run ruff check / ruff format --diff clean
• prek run on the changed files — all hooks pass

🤖 Generated with Claude Code

https://claude.ai/code/session_011iv7L8vX7KgrAukeeXKfa6

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 68.09%. Comparing base (c1391db) to head (3d238a7).

Additional details and impacted files

@@           Coverage Diff           @@
##            jazzy     #397   +/-   ##
=======================================
  Coverage   68.09%   68.09%           
=======================================
  Files         102      102           
  Lines        6320     6320           
=======================================
  Hits         4303     4303           
  Misses       2017     2017           

Flag	Coverage Δ
cpp-jazzy	68.09% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.