Don't force reprojection when it is possible in the CLI accuracy workflow

[marinebcht]

[Work In Progress]

Resolves #909

sampling_grid = None can be possible:

• coreg with ICP and CPD (work even if datasets are not reprojected)
  • ICP/CPD methods only: OK
  • Other method(s) only: KO
  • ICP/CPD + other method(s): KO
• no coreg : KO

Tests ne no resampling with :

• overlapping dems: dem vs smaller dem
• no overlapping dems ?

/!\ Openned GlacioHack/geoutils#900

[marinebcht]

Hey ! I found out that in even in ICP/CPD, there is a always a reprojection between ref and tba here

xdem/xdem/coreg/base.py

Line 156 in 7d84904

dem_to_be_aligned = dem_to_be_aligned.reproject(reference_dem, silent=True)

(in the _preprocess_coreg_fit_raster_raster function). Am I correct ? @rhugonnet (Checked #759 in case) Perhaps we should start there before working on the workflow extention @belletva

[belletva]

Hey ! I found out that in even in ICP/CPD, there is a always a reprojection between ref and tba here

xdem/xdem/coreg/base.py

Line 156 in 7d84904

dem_to_be_aligned = dem_to_be_aligned.reproject(reference_dem, silent=True)

(in the _preprocess_coreg_fit_raster_raster function). Am I correct ? @rhugonnet (Checked #759 in case) Perhaps we should start there before working on the workflow extention @belletva

good catch!

[rhugonnet]

It is not forced in #759, with edits adding a reproj_same_grid argument:

https://github.com/rhugonnet/xdem/blob/5c9d8c3211db3024425135904c57fe11317f4245/xdem/coreg/base.py#L189

[marinebcht]

Oki top @rhugonnet read your PR #759 a little too quickly ! :)
I think I need to wait that this PR is finished @belletva