feat: Slot-basierter Auto-Refresh mit konfigurierbarer Startzeit

Auto-Refresh nutzt jetzt eine feste Anker-Uhrzeit (refresh_start_time) statt
reinem Intervall-basiertem Driften. Verpasste Slots werden max. 1x aufgeholt.
Bestehendes Intervall-Verhalten bleibt als Fallback erhalten (ohne Startzeit).

Migration: Bestehende Auto-Lagen erhalten 07:00 als Startzeit.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/models.py

@@ -49,6 +49,7 @@ class IncidentCreate(BaseModel):
     type: str = Field(default="adhoc", pattern="^(adhoc|research)$")
     refresh_mode: str = Field(default="manual", pattern="^(manual|auto)$")
     refresh_interval: int = Field(default=15, ge=10, le=10080)
+    refresh_start_time: Optional[str] = Field(default=None, pattern=r"^([01]\d|2[0-3]):[0-5]\d$")
     retention_days: int = Field(default=0, ge=0, le=999)
     international_sources: bool = True
     include_telegram: bool = False
@@ -62,6 +63,7 @@ class IncidentUpdate(BaseModel):
     status: Optional[str] = Field(default=None, pattern="^(active|archived)$")
     refresh_mode: Optional[str] = Field(default=None, pattern="^(manual|auto)$")
     refresh_interval: Optional[int] = Field(default=None, ge=10, le=10080)
+    refresh_start_time: Optional[str] = Field(default=None, pattern=r"^([01]\d|2[0-3]):[0-5]\d$")
     retention_days: Optional[int] = Field(default=None, ge=0, le=999)
     international_sources: Optional[bool] = None
     include_telegram: Optional[bool] = None
@@ -82,6 +84,7 @@ class IncidentResponse(BaseModel):
     status: str
     refresh_mode: str
     refresh_interval: int
+    refresh_start_time: Optional[str] = None
     retention_days: int
     visibility: str = "public"
     summary: Optional[str]