fix: split inner-loop yi_max clamp to fix HOLD-state windup overshoot

Renames pid_hold/pid_heat/pid_cool to pid_outer/pid_inner/pid_inner_cool
to match what actually runs when, and splits inner-loop config into
Inner.Heat/Inner.Hold/Inner.Cool so the same PID instance gets a tight
yi_max ceiling only while HOLD drives it, without capping legitimate
1.5 K/min ramps. Fixes the overshoot from docs/overshoot_hold_windup.md
where a cold-water disturbance during HOLD wound up pid_heat's integral
term with no anti-windup engagement, taking ~35s+ to unwind naturally.

Breaking config change: Hold/Heat/Cool -> Outer/Inner.{Heat,Hold,Cool}
in config.json, both .tpl templates, the pid/sud demo scripts, and
replay_sim.py's CLI flags. Adds tests/components/pid/ (stdlib unittest)
covering the Pid clamp/recovery behavior and closed-loop disturbance,
ramp, and HOLD<->HEAT transition cases.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DGQhVQ2Y3yXAQTXhrxVd5u

components/sud_forecast.py

@@ -46,7 +46,7 @@ class SudForecastEstimator:
 	estimate() also duty-cycles the PID's continuous output across the
 	heater's own discrete power steps (see its actuate() closure), exactly
 	like the real run's HeaterTask/device chain does - feeding tc.get_power()
-	straight to the plant instead lets pid_heat's own oscillatory tendency
+	straight to the plant instead lets pid_inner's own oscillatory tendency
 	reach the plant undamped, producing a forecast far more jagged than any
 	real run actually is (the discrete steps end up duty-cycling it back
 	down to something close to the commanded average)."""