Scraping a URL with timestamp & refresh

Configuration
1

Hi!
I’m trying to get some data (tram departures, specifically) from a webpage using the scrape component. I use the following configuration:

scrape:
  - resource_template: <base URL>{{now().timestamp() | timestamp_custom("%d.%m.%Y&itdTime=%H%%3A%M")}}
    method: POST
    timeout: 20
    scan_interval: 3600
    sensor:
      - name: "tram_departure_0"
        unique_id: "tram_departure_0"
        select: <css selector>

where I omitted the fixed part of the URL outside the template-part, the specific css selector and a few more sensors for the same page as they do not matter here.

This seems to work in principle: when I start homeassistant, the sensors are filled with the correct data. However, it appears that the date-part for the URL never changes.

Is this due to the handling of now()? Or is the template just evaluated once on startup instead of every refresh of the sensor? (They do get refreshed every hour, just with the old timestamp in the URL.)

Any idea how to solve that problem?

Side note: is it possible to disable auto-refresh altogether and just use
homeassistant.update_entity as needed for sensors declared in the configuration file?