Gordias fails to save large files with iterative flag

When creating the temporary data array for the placeholder data, iris fails to save the array in the temporary file because it is to big. E.g. shape:(47482, 1801, 3600). Results in "numpy.core._exceptions._ArrayMemoryError: Unable to allocate 1.12 TiB for an array with shape (47482, 1801, 3600) and data type float32"

We should test to use dask broadcast instead of numpy and also to rechunk the temporary data before we save it.