Get background loading/some block parallelization back

After #115 was implemented to clean up the strided block mapping, the background loading was temporarily removed:

Lines 168 to 174 in 108234d

    
           # block_gen = vrt.iter_blocks( 
        
           #     block_shape=block_shape, 
        
           #     overlaps=overlaps, 
        
           #     skip_empty=True, 
        
           #     nodata_mask=nodata_mask, 
        
           #     show_progress=show_progress, 
        
           # )

We should either add it back in as it was, or implemented a better way to run the body of the for-loop on multiple blocks so that one can be loading data while the other is processing.

Added back in #201

	# block_gen = vrt.iter_blocks(
	# block_shape=block_shape,
	# overlaps=overlaps,
	# skip_empty=True,
	# nodata_mask=nodata_mask,
	# show_progress=show_progress,
	# )