Combine rasters with same extent, but different resolution

[edornd]

I have a time series of Sentinel-2 images, where each band is stored in its own file. I'd like to combine bands and time to create a multidimensional array, and store it as a single HDF5/NetCDF file for convenience.
I managed to use open_mfdataset to create this tensor, however the files with a lower resolution do not get interpolated, but rather "stretched" with holes filled by NaNs:

data = xr.open_mfdataset(
    images,
    concat_dim=["time", "band"],
    chunks="auto",
    engine="rasterio",
    combine="nested",
    coords="minimal",
    compat="override",
)

gives the following Dataset:

<xarray.Dataset>
Dimensions:      (x: 264, y: 264, band: 12, time: 4)
Coordinates:
  * x            (x) float64 30.06 30.06 30.06 30.06 ... 30.08 30.08 30.09 30.09
  * y            (y) float64 13.97 13.97 13.97 13.97 ... 13.99 13.99 13.99 13.99
  * band         (band) int64 1 1 1 1 1 1 1 1 1 1 1 1
    spatial_ref  int64 ...
Dimensions without coordinates: time
Data variables:
    band_data    (time, band, y, x) float32 dask.array<chunksize=(1, 1, 264, 264), meta=np.ndarray>

However some of the bands have NaNs due to the stretch.

I tried to interpolate the lower resolution bands in the preprocess function: the resulting xarray Dataset is in fact now 264x264 (which is my desired shape), however the final output of the open_mfdataset now becomes way bigger than expected:

<xarray.Dataset>
Dimensions:      (x: 787, y: 786, band: 12, time: 4)
Coordinates:
  * x            (x) float64 30.06 30.06 30.06 30.06 ... 30.08 30.08 30.09 30.09
  * y            (y) float64 13.97 13.97 13.97 13.97 ... 13.99 13.99 13.99 13.99
  * band         (band) int64 1 1 1 1 1 1 1 1 1 1 1 1
    spatial_ref  int64 ...
Dimensions without coordinates: time
Data variables:
    band_data    (time, band, y, x) float32 dask.array<chunksize=(1, 1, 786, 787), meta=np.ndarray>

Is it possible to obtain a Dataset with shape (264, 264), where the lower resolution bands are interpolated to the desired dimension?
Thanks!