adding gradientdescending coreg

[liuh886]

New coreg method ! faster and support point coregistration.

what is is about, and how to use it: Zhihao_test_on_new_coreg.ipynb

Changes Made

coreg.py

• add new subclass: 'xdem.coreg.GradientDescending'
• edit 'xdem.coreg.fit' to support DataFrame as a ref_dem in coreg.
• add function 'apply_xyz_shift_df' to shift DataFrame.
• add subset_gdf to subset DataFrame, such as ICESat-2 measurements.
• add residuals_df to Calculate the difference between the DEM and points (a dataframe has 'E','N','Z') after applying shift.
• add df_sampling_from_dem to generate a datafram from a dem by random sampling.

environment.xml

• add noisyopt. This is a gradient descending implement.

data

• add a test dataset 'dtm_10_snowfree.tif' and 'icesat_2_snowfree_qc.csv'

test_coreg.py

• add test_coreg_example_shift_test or comparison of two coreg algorithms.
• add test_coreg_example_gradiendescending for further test

How to Test

• Do tests on DEMs, using it like NuthKaab.
• Do tests on ICESat-2. The point measurement should be stored at a DataFrame, and has columns 'N','E' (in same CRS with DEM), or 'lattitude','longitude' in CRS 4326. You need to load_gdf() to clip it or do it by yourself before sending it to 'xdem.coreg.GradientDescending'.

Be aware of

• This method is still under development and further testing is required before it can be used in academic projects.