janno contamination consistency ignores NA

[stschiff]

The current implementation complains if NA is in the Contamination column, with no values in the two other required Contamination fields. That is obviously a bug, because the specification makes no semantic difference between an empty string and NA.

[nevrome]

Hm... I can not reproduce this. How exactly did you encode NA? Literally with NA or with the correct n/a? Maybe that's already the issue? Or how could I reproduce this behaviour?

Independent of that we should certainly improve the validation for these sets of columns. We already discussed this in #153 and I still think that's the right way to approach this. Also considering semantic changes we may want to apply to the schema (poseidon-framework/poseidon-schema#64).

[stschiff]

Yeah, very sorry to have wasted your time 🙈, I used NA. Good learning session for me. I'll close this.