fix(output): a HoldChrg row reports the limit it holds at (#4445)

Now that HoldChrg is decided from charge_limit_best, quoting the
predicted end-of-window SoC in its reason could contradict the row, for
example 'already at or above the 70% target' on a row at 55% when PV is
predicted to carry the SoC past a 50% limit. Show the limit instead,
which matches main whenever the prediction equals the limit (the usual
hold case). Chrg rows keep showing the predicted SoC.

Tests cover the hold level and the PV-overshoot case (prediction above
the limit), which now shows HoldChrg at the limit, not Chrg.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: chalfontchubby

apps/predbat/output.py

@@ -1423,10 +1423,11 @@ def import_rate_color(rate):
                 if "target" in self.charge_window_best[charge_window_n]:
                     target = self.charge_window_best[charge_window_n]["target"]
 
-                # Classify Chrg vs HoldChrg against the real instruction (charge_limit_best), but display the
-                # window's predicted end-of-window SoC ("target", from plan.py), which is what the Limit column
-                # documents. Classifying against the prediction showed a full-rate charge that can't finish
-                # within the window as HoldChrg whenever the row's SoC sat above that prediction (issue #4445).
+                # Classify Chrg vs HoldChrg against the real instruction (charge_limit_best), but have a Chrg row
+                # display the window's predicted end-of-window SoC ("target", from plan.py), which is what the
+                # Limit column documents. Classifying against the prediction showed a full-rate charge that can't
+                # finish within the window as HoldChrg whenever the row's SoC sat above that prediction (issue #4445).
+                # A HoldChrg row shows the limit it is held at, which the prediction need not match.
                 instruction_percent = calc_percent_limit(limit, self.soc_max)
                 limit_percent = calc_percent_limit(target, self.soc_max)
                 if limit > 0.0:
@@ -1440,6 +1441,7 @@ def import_rate_color(rate):
                         state = "HoldChrg→"
                         state_color = "#34DBEB"
                         raw_state = "HoldChrg"
+                        limit_percent = instruction_percent
                         reason_parts.append({"code": "hold_charge_at_target", "params": {"target_percent": limit_percent}})
                     else:
                         state = "Chrg↗"
@@ -1459,7 +1461,7 @@ def import_rate_color(rate):
                     show_limit = str(limit_percent)
                     had_state = True
                     if plan_debug:
-                        show_limit += " ({})".format(str(calc_percent_limit(limit, self.soc_max)))
+                        show_limit += " ({})".format(str(instruction_percent))
                     raw_state_target = str(limit_percent)
             else:
                 if export_window_n >= 0:

apps/predbat/tests/test_plan_why_reason.py

@@ -450,13 +450,31 @@ def render(car_hold_minutes=None):
         print("ERROR: #4445 plan_debug Limit cell should read '60 (80)' (predicted SoC, then limit)")
         failed = True
 
-    # A real hold still classifies as HoldChrg: the SoC is at or above the real limit
+    # A real hold still classifies as HoldChrg (the SoC is at or above the real limit), and reports the
+    # 80% limit it is held at rather than the predicted 60%
     my_predbat.predict_soc_best = _flat_soc(my_predbat, 9.0)  # 90%, above the 80% limit
     _, raw_plan = render()
     row = _get_row(raw_plan, minutes_now)
     if row is None or _codes(row) != ["hold_charge_at_target"]:
         print("ERROR: #4445 SoC above the real limit should still show HoldChrg: {}".format(row and _codes(row)))
         failed = True
+    elif row["reasons"][0]["params"] != {"target_percent": 80} or row["state_target"] != "80":
+        print("ERROR: #4445 HoldChrg should report the 80% limit it holds at: {} -> {}".format(row["reasons"][0]["params"], row["state_target"]))
+        failed = True
+
+    # The prediction can also sit above the limit (PV carries the SoC past it): a row already above the
+    # real limit is held there, not charged towards the predicted SoC
+    my_predbat.charge_window_best[0]["target"] = 9.5  # predicted to reach 95% by window end
+    my_predbat.predict_soc_best = _flat_soc(my_predbat, 8.5)  # 85%, between the 80% limit and the 95% target
+    _, raw_plan = render()
+    row = _get_row(raw_plan, minutes_now)
+    if row is None or _codes(row) != ["hold_charge_at_target"]:
+        print("ERROR: #4445 SoC above the limit but below a higher prediction should show HoldChrg: {}".format(row and _codes(row)))
+        failed = True
+    elif row["reasons"][0]["params"] != {"target_percent": 80} or row["state_target"] != "80":
+        print("ERROR: #4445 HoldChrg below a higher prediction should report the 80% limit: {} -> {}".format(row["reasons"][0]["params"], row["state_target"]))
+        failed = True
+    my_predbat.charge_window_best[0]["target"] = 6.0  # back to the 60% prediction for the text summary check below
 
     # --- Test 2c: get_charge_export_text reports the predicted SoC too, falling back to the limit ---
     print("Test get_charge_export_text shows the predicted SoC, or the limit when there is no prediction")