Deux voies équivalentes pour obtenir les données de prix et de CO2 dans Home Assistant

type: custom:apexcharts-card
header:
  show: true
  title: Energy Price Forecast EU
graph_span: 3d
span:
  start: day
now:
  show: true
  label: Maintenant
yaxis:
  - id: price
    decimals: 3
    apex_config:
      title:
        text: EUR/kWh
all_series_config:
  yaxis_id: price
  type: line
  curve: stepline
  extend_to: false
  stroke_width: 2
  float_precision: 4
  show:
    legend_value: false
series:
  # Replace ALL three placeholders with the SAME price or retail sensor.
  - entity: sensor.CHANGE_ME
    name: Officiel (day-ahead)
    color: "#43a047"
    data_generator: |
      const raw = [...(entity.attributes.raw_today ?? []),
                   ...(entity.attributes.raw_tomorrow ?? [])];
      const slots = raw.map(e => ({
        start: Date.parse(e.start), end: Date.parse(e.end), value: e.value
      })).filter(e => Number.isFinite(e.start) && Number.isFinite(e.end)
        && e.end > e.start && typeof e.value === 'number' && Number.isFinite(e.value))
        .sort((a, b) => a.start - b.start);
      const points = [];
      let previousEnd = null;
      for (const slot of slots) {
        if (previousEnd !== null && slot.start > previousEnd) {
          points.push([previousEnd, null], [slot.start - 1, null]);
        }
        points.push([slot.start, slot.value], [slot.end - 1, slot.value]);
        previousEnd = slot.end;
      }
      return points;
  - entity: sensor.CHANGE_ME
    name: Prévision actuelle
    color: "#fb8c00"
    data_generator: |
      const raw = entity.attributes.raw_forecast ?? [];
      const slots = raw.map(e => ({
        start: Date.parse(e.start), end: Date.parse(e.end), value: e.value
      })).filter(e => Number.isFinite(e.start) && Number.isFinite(e.end)
        && e.end > e.start && typeof e.value === 'number' && Number.isFinite(e.value))
        .sort((a, b) => a.start - b.start);
      const points = [];
      let previousEnd = null;
      for (const slot of slots) {
        if (previousEnd !== null && slot.start > previousEnd) {
          points.push([previousEnd, null], [slot.start - 1, null]);
        }
        points.push([slot.start, slot.value], [slot.end - 1, slot.value]);
        previousEnd = slot.end;
      }
      return points;
  - entity: sensor.CHANGE_ME
    name: Prévision enregistrée (première reçue)
    color: "#1e88e5"
    stroke_dash: 5
    data_generator: |
      const raw = entity.attributes.raw_forecast_reference ?? [];
      const slots = raw.map(e => ({
        start: Date.parse(e.start), end: Date.parse(e.end), value: e.value
      })).filter(e => Number.isFinite(e.start) && Number.isFinite(e.end)
        && e.end > e.start && typeof e.value === 'number' && Number.isFinite(e.value))
        .sort((a, b) => a.start - b.start);
      const points = [];
      let previousEnd = null;
      for (const slot of slots) {
        if (previousEnd !== null && slot.start > previousEnd) {
          points.push([previousEnd, null], [slot.start - 1, null]);
        }
        points.push([slot.start, slot.value], [slot.end - 1, slot.value]);
        previousEnd = slot.end;
      }
      return points;

Help me tailor an ApexCharts card for Energy Price Forecast EU. First ask for my exact entity ID, sensor unit, integration version, hourly/quarter-hourly settlement and whether I want market or retail prices. Do not ask for an API key.
From integration 1.12.0, Price forecast series and Current retail price expose:
- raw_today and raw_tomorrow: published day-ahead slots.
- raw_forecast: current forecast slots beyond official coverage.
These lists contain {start, end, value}.
- raw_forecast_reference: {start, end, value, captured_at}; the first forecast received for a then-future slot, saved locally. Later forecasts and official prices never overwrite it.
References accumulate only after updating, survive restarts and are removed after the target market day ends. captured_at is the HA observation time, not model creation.
Use green for official prices, orange for the current forecast and dashed blue for the saved forecast, on the same axis in the sensor's unit. Require ApexCharts Card 2.1.0 or newer. Use each series' data_generator and extend_to: false. Draw start through end, preserve gaps with null, and retain genuine zero/negative prices. Never copy official values into a forecast line. Use only the documented attributes.
Optionally ask about the hourly/weekend plan: its hours attribute contains {start, end, average_value}. Mark those times separately without changing the plan. Return complete YAML for a manual Lovelace card and explain where to paste it. A chart alone does not switch a charger.

Answer in French.

Help me build a safe Home Assistant automation using the Energy Price Forecast EU API.

Read the API contract first:
https://energypriceforecast.eu/openapi/integration-api.json

Use /api/v1/home-assistant/summary and do not invent fields. The API uses official day-ahead prices wherever available and forecast values only for the remaining open horizon.

Before writing YAML, ask me no more than these questions, one at a time:
1. Country or bidding zone?
2. Which load should be controlled and which Home Assistant entity switches it?
3. Power, required runtime and latest completion time?
4. Optimise for price, CO2 or both?
5. Free without a key and up to 48 hours, or Private Pro with up to 120 hours?
6. Do I use packages or configuration.yaml?

Rules for your result:
- Free: hours=48 and no Authorization header.
- Private Pro: hours=120 and header Authorization: Bearer YOUR_API_KEY.
- Never ask me to paste the real key into the chat.
- Handle HTTP errors and check meta.api_key_state and meta.allowed_horizon_hours.
- Briefly explain every API property you use.
- Produce complete, correctly indented YAML and state exactly where it belongs.
- Start with a safe test mode using a notification or input_boolean. Put real load switching in a separate, clearly marked step.
- Define safe behaviour for stale, missing or implausible data. An API error must never switch the load on unexpectedly.
- Finish with a checklist: HTTP 200, sensors populated, allowed horizon correct and test mode triggered correctly.

Ask when information is missing. Do not silently assume entity names, tariffs or electrical limits.

Answer in French.