Stefan Verhoeven

This moves the sorting of scores back into the mathms library which was difficult because we don't know what a similarity function returns.

You could use `rdkit.Chem.Atom('C').GetMass()` to calculate mass or use https://pypi.org/project/molmass/

See https://numpy.org/doc/stable/reference/generated/numpy.ndarray.flags.html how to mark numpy arrays as read only.

The Spikes.clone() should still copy, but Spikes.peaks and Spikes.losses should return a read only array.

Keep the CONTRIBUTING guide it answers questions for potential developers. The release instructions are for a core matchms developer. Maybe README is also not the best place as release instructions...

Doh, Instructions on https://anaconda.org/bioconda/matchms also don't because rdkit must come from conda-forge channel. Yes, lets drop nlesc channel 1. Remove all the anaconda build stuff from CI workflows and docs...

I am trying to install the `matchms-0.8.1-pyh3252c3a_1.tar.bz2` from bioconda, but it is failing with ```shell conda install -c bioconda matchms=0.8.1=pyh3252c3a_1 Collecting package metadata (current_repodata.json): done Solving environment: failed with initial...

Where make a lot of numpy copies we could switch to i[mmutable numpy](https://numpy.org/doc/stable/reference/generated/numpy.ndarray.flags.html) arrays and only copy when asked for . (update by @jspaaks): please continue the discussion about this...

Add `is_symmetrical=False` argument to calculate_scores to not calculate lower triangle of matrix. (update by @jspaaks): please continue the discussion about this aspect in https://github.com/matchms/matchms/issues/59

Before we work on this issue we should have consumers of the KNIME workflow and nodes