model: TroCandidate statt OpenPoint, tro_overrides nutzt model-Klassen (Schema v3)

- OpenPoint -> model.TroCandidate (type, position, belongsTo, note);
  TroDocument.candidates ersetzt openPoints, Schema sps_skel-tro-v3
- Positionsbehandlung von TroBase und TroCandidate im Mixin _Positioned
- tro_overrides.py: eigene ConnectionSpec/OpenPoint entfallen, stattdessen
  model.Connection / model.TroCandidate; einheitliche Feldnamen
  (tro_id/separator, hosts als Liste)
- Fixtures, Beispiel-JSON und Doku auf v3 nachgezogen

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

tests/test_model.py

@@ -6,7 +6,7 @@ import json
 import pytest
 from pydantic import ValidationError
 
-from model import SCHEMA, Analysis, Connection, DerivedTro, TroDocument
+from model import SCHEMA, Analysis, Connection, DerivedTro, TroCandidate, TroDocument
 from scl_skeleton import render
 
 
@@ -64,7 +64,18 @@ def test_json_uses_tro_json_keys(document):
     assert first["items"] == {"Separator": 1, "Switch": 1}
     assert first["position"] == {"x": 1030.0, "y": 4700.0, "z": 1500.0}
     assert data["connections"][1] == {"from": "TRO01", "to": "TRO02", "kind": "weiche"}
-    assert "openPoints" in data
+    assert data["candidates"] == []
+
+
+def test_candidate_round_trip():
+    candidate = TroCandidate(name="Qualitaetskontrolle", type_name="1Sep1Swi",
+                             position=(-1026.0, 5250.05, 1478.5), belongs_to="TRO09")
+    data = json.loads(candidate.model_dump_json(by_alias=True))
+    assert data["type"] == "1Sep1Swi"
+    assert data["position"] == {"x": -1026.0, "y": 5250.05, "z": 1478.5}
+    assert data["belongsTo"] == "TRO09"
+    assert TroCandidate.model_validate(data) == candidate
+    assert candidate.describe_position() == "-1026, 5250"
 
 
 def test_items_and_fb_block_follow_the_type():