Improve resample_in_space and rectify_dataset

[konstntokas]

This PR improves the resampling_in_space method, if another CRS is required and therefore a reprojection needs to be done.

• The method xcube.core.GridMapping.transform now supports lazy execution. If
  computations based on actual data are required—such as determining whether the
  grid mapping is regular or estimating the resolution in the x or y direction—only a
  single chunk is accessed whenever possible, ensuring faster performance.
  -> see large example examples/notebooks/resampling/reproject_large_esa_cci_landcover.ipynb execution time of cell 5 drops from 1.42min to ~800ms.

• The function xcube.core.resampling.rectify_dataset now supports xarray.Datasets
  containing multi-dimensional data variables structured as var(..., y_dim, x_dim).
  The two spatial dimensions (y_dim and x_dim) must occupy the last two positions
  in the variable's dimensions.
  -> see large example examples/notebooks/resampling/reproject_large_esa_cci_landcover.ipynb execution time of cell 6 drops from 4.51min to ~2.16ms, even though now 5 time steps are selected compared to just one.

Checklist:

• Add unit tests and/or doctests in docstrings
• Add docstrings and API docs for any new/modified user-facing classes and functions
• New/modified features documented in docs/source/*
• Changes documented in CHANGES.md
• GitHub CI passes
• AppVeyor CI passes
• Test coverage remains or increases (target 100%)

[konstntokas]

CI fails due to two tests failures in test/core/gen2/test_progress, which is not related to this PR.

[forman]

Great, thanks! Please take care of my suggestions, then merge.