Implement ForecastIndex (an xarray custom index)

As flagged in #47, #38 and #33, the long awaited flexible/custom index functionality is now available in xarray (see the [v2025.07.1 release notes](https://docs.xarray.dev/en/stable/whats-new.html#v2025-07-1-july-09-2025)). See the [SciPy presentation](https://www.youtube.com/watch?v=I-NHCuLhRjY) and [custom index gallery](https://xarray-indexes.readthedocs.io/index.html) for details.

There is already a custom index for handling forecast data - the `rolodex.ForecastIndex` described at [this gallery page](https://xarray-indexes.readthedocs.io/earth/forecast.html).

The gallery page talks about the fact that there are four different ways of constructing views of forecast data (or more specifically, of “Forecast Model Run Collections; FMRCs"). By using the `rolodex.ForecastIndex` we can efficiently select and slice data along any of those four views, essentially doing away the cumbersome functions we've written in our `array_handling.py` module for swtiching views of the data.

Start by reading array_handling.py and the rolodex.ForecastIndex documentation linked in the issue, along with xarray's custom-index release notes. Compare the existing view-switching functions with the four FMRC views described in the gallery. Done means ForecastIndex provides the intended selection and slicing views so those cumbersome functions can be replaced.

Written by the indexing model from the issue text.