Release 0.2.6

[atrabattoni]

🚀 We are excited to announce the release of Xdas 0.2.6!

A big thank you to all contributors for the many new features, improvements, and fixes included in this release 🙌

Happy coding 💻
— The Xdas team

New features

• Add xdas.open that automatically infers which xdas.open_* function to use (@atrabattoni, @yetinam).
• Add automatic engine detection to every xdas.open_* function (@atrabattoni, @yetinam).
• Add pathlib.Path support as input for all xdas file-related functions and methods (@atrabattoni).
• Add xdas.io.compressed that compresses a specific dataset in an HDF5 file while preserving the rest of the file structure and metadata (@marbail).
• Add xdas.concat_coords to merge coordinates. Also Coordinate.append is now Coordinate.concat (avoid in-place confusion) and xdas.concatenate has now a preferred alias xdas.concat (@atrabattoni).

Improvements

• All to_netcdf methods now have a create_dirs argument to create intermediate directories if necessary (@aurelienfalco).
• Make DataArray.sel handle overlaps when slicing, and xdas.split can split on overlaps or gaps now (@atrabattoni).
• New io.Engine backend system to register different file formats (@atrabattoni).
• Make open_mfdataarray raise RuntimeError when opening all files fails (@asladen).
• Add "prodml" engine (@atrabattoni) and make "optasense" and "sintela" aliases of it (@atrabattoni).
• Add the component_strategy argument to the xdas.atoms.MLPicker to choose whether to use the same component on the 3 channels or to use one channel and set the others to 0 based (@marbail).
• Make DataArray.rename capable of renaming dims and coords (@atrabattoni).
• Add parallel argument to most open* functions to let the user choose the file opening strategy (@atrabattoni).

Bugs Fixed

• Fix memory accumulation when slicing multiple times data arrays, e.g. when using atoms (@atrabattoni).
• Fix non-terminating loaders and writers in xdas.processing (@atrabattoni).
• Fix/improve distance handling for: "apsensing", "febus", "optasense", "silixa", and "sintela" (@atrabattoni).
• Add dim swapping handling for the "prodml" based engines with the swapped_dims=False kwarg (@atrabattoni).
• Fix ASN ROI handling (@asladen).
• Use the annotate_batch_pre model's function to normalize in xdas.atoms.MLPicker (@marbail).
• Fix the RuntimeError encountered when using open_mf* functions in scripts due to the use of multiprocessing by using the loky library (@atrabattoni)