…st one

+ objects.cluster.split_box2sub_boxes(): if the last step is too small (<5% or <5 pixels), merge it to the 2nd last one. This change is necessary to avoid the reshaping error when the step size of the last sub-box is 1, turning a 3D matrix into 2D, thus, resulting in a reshape error, such as in diff.py.

+ update the num_box based on the returned box_list from split_box2sub_boxes(), as it maybe different from the initially input num_box. This is similar to insarlab#940, but search and update all the other instances.

…` for tiny last step (#1024)

+ `save_gdal.py`: convert numpy array from boolean to uint8, as GDAL does not support boolean data type, such as used in all the mask files.

+ `objects.cluster.split_box2sub_boxes()`: avoid the tiny last step (<5% of the normal step, or <5 pixels), by merging it into the 2nd last one. This happens to very large datasets sometimes. For example, if the last step has a size of 1, a 3D matrix will be turned into 2D, resulting in a reshape error.

+ `objects.cluster.split_box2sub_boxes()`: return the final number of sub_boxes, in addition to the current list of sub_boxes, for easier usage (as it may be different from the initial num_box as identified in #940). Update the usage in all occurrences:
   - objects.resample
   - dem_error
   - diff
   - reference_date
   - timeseries2velocity

…st one

+ objects.cluster.split_box2sub_boxes(): if the last step is too small (<5% or <5 pixels), merge it to the 2nd last one. This change is necessary to avoid the reshaping error when the step size of the last sub-box is 1, turning a 3D matrix into 2D, thus, resulting in a reshape error, such as in diff.py.

+ update the num_box based on the returned box_list from split_box2sub_boxes(), as it maybe different from the initially input num_box. This is similar to insarlab#940, but search and update all the other instances.

…` for tiny last step (insarlab#1024)

+ `save_gdal.py`: convert numpy array from boolean to uint8, as GDAL does not support boolean data type, such as used in all the mask files.

+ `objects.cluster.split_box2sub_boxes()`: avoid the tiny last step (<5% of the normal step, or <5 pixels), by merging it into the 2nd last one. This happens to very large datasets sometimes. For example, if the last step has a size of 1, a 3D matrix will be turned into 2D, resulting in a reshape error.

+ `objects.cluster.split_box2sub_boxes()`: return the final number of sub_boxes, in addition to the current list of sub_boxes, for easier usage (as it may be different from the initial num_box as identified in insarlab#940). Update the usage in all occurrences:
   - objects.resample
   - dem_error
   - diff
   - reference_date
   - timeseries2velocity