[zebraengine]

Problem

PyPI serves 0.5.0; main is eight PRs past it. A new install still gets a drift watch that can fire on its own amp controller's caps, an I² current law this install's data rejects, and no calibration probe.

Code touched

pyproject.toml: 0.5.0 -> 0.6.0; uv.lock to match. Nothing else; the release content is what already merged:

• fix: estimate heat-rise drift by regression, and only from free-running windows #29: heat-rise drift estimated by regression over days, ambient and current, from free-running windows only.
• feat: automate a monthly calibration probe, and stop discarding its fits #30: automated calibration probe in the amp controller (--probe-amps), and its fits no longer discarded.
• feat: fit how rise scales with current, and restore straight to the sustainable rate #31: the current exponent is fitted per install instead of assumed 2; forecasts report sustainable_max_a; the controller restores straight to it.
• fix: don't end a steady-current run on a ramp-up overshoot #32: a ramp-up overshoot in the first minute no longer ends a steady-current run.
• feat: backtest the forecast against every recorded session, with its sampling bias controlled #35: contrib/backtest_forecast.py, scoring the forecast against the whole history with sampling bias controlled.
• feat: a calibration probe plan — two currents, each with a cold cable and a warm one #36: probe plan: several currents, each with a cold cable and a warm one.
• fix: probe plan survives a reboot #37: controller state moves out of /tmp; probe history is rebuilt from the event log.
• feat: a settle-on-target amp controller and a closed-loop backtest to judge it (experimental, not wired in) #38: experimental settle-on-target controller and closed-loop backtest in contrib/, not wired in.

Risk

Minor bump: additive API fields and new contrib/ flags; no schema migration. Fits are recomputed rather than stored, so on first start after upgrade the rise medians and the drift verdict can shift as the fitted current law and the free-plateau gate apply to existing history. Merging this PR publishes nothing; the tag does.

Verification

• uv run pytest on main at a2552db: 176 passed. uv lock --check passes with the bumped lockfile.
• Each included PR carried its own verification.
• After publish: PyPI JSON and uvx --from wallmonitor==0.6.0 wallmonitor --help.

Deploy: tag v0.6.0 on the merge commit after merge; GitHub Release created published.

🤖 Generated with Claude Code