Simplify loading NetCDF files in `load_solar_pv_data`

[JackKelly]

Detailed Description

It used to be necessary to do this to load NetCDF quickly from a cloud storage bucket:

  with fsspec.open(filename, mode="rb") as file:
      file_bytes = file.read()

  with io.BytesIO(file_bytes) as file:
      pv_power = xr.open_dataset(file, engine="h5netcdf")
      pv_power = pv_power.sel(datetime=slice(start_dt, end_dt))
      pv_power_df = pv_power.to_dataframe()

  # Save memory
  del file_bytes

But it looks like the underlying libraries have been optimised. Now, it appears that the following, simpler version of the code is at least as fast (I tested on Google Cloud today):

  with fsspec.open(filename, mode="rb") as file:
      pv_power = xr.open_dataset(file, engine="h5netcdf")
      pv_power = pv_power.sel(datetime=slice(start_dt, end_dt))
      pv_power_df = pv_power.to_dataframe()

As such, we can simplify this code in nowcasting_dataset.data_sources.pv.pv_data_source.load_solar_pv_data.