fix: make dataset nodes ordering stable

[tristan-f-r]

This is the main change:

spras/spras/dataset.py

Lines 107 to 108 in a61bfce

	# We get uniqueness afterwards to make `load_files_from_dict` have a well-defined node ordering
	nodes = list(pd.concat([self.interactome.Interactor1, self.interactome.Interactor2]).unique())

Some algorithms depend on the ordering of their nodes, which is why this is marked as needed for benchmarking: we can guarantee immutable generated outputs this way. We never want the behavior of #362 falsely flagged as immutable.

---

This also adds tests to all inputs, closing #336.

This will be annoying to review. I have at least gone through and compared it with the regular input file to see that this looks right, but a thorough review of all of the test files will be unfortunately tedious.

[read-the-docs-community]

Documentation build overview

📚 spras | 🛠️ build #29151356 (fd804a2) | 🔍 preview

Files changed

Comparing with latest (a9cff01...fd804a2)

Show files (4) | 4 modified | 0 added | 0 deleted

File	Status
genindex.html	📝 modified
contributing/index.html	📝 modified
fordevs/modules.html	📝 modified
fordevs/spras.html	📝 modified