Fixing tests

[lassedochreden]

relates to arm64 build errors on BioC

Fixed the error for countInteractions() - seems to be related to small floating point errors

Did not fix the error in test_integration_interactions.R - did not find a quick fix - potentially related to different NA handling (see screenshot below) - would be great, if @nilseling could have a look.

Will update NEWS and DESCRIPTION upon approval.

[nilseling]

Thanks for looking into it! Could you figure out which p value is correct by comparing the count against the permutations? imcRtools let's you set the return_samples parameter for this.

[nilseling]

Ok, in this case I'm 95% sure that the old implementation of the function is generating a wrong p value due to floating point issues. Would be good to double check though

[lassedochreden]

Also now fixed test_integration_interactions.R - Was related to machine precision issues, which we did not consider when manually calculating the p-values (despite doing so in the function implementation).

Will merge, if you agree.

[lassedochreden]

@SchulzDan - Please push to BioC, when you can.

[nilseling]

I noticed that there is this max(ct_obs)==0 call in utils.R when calculating the p values. And since a count of 0 is now not detected as 0 anymore I wondered if this also needs to be replaced.