use limma's nonEstimable and is.fullrank

[sreichl]

Consider replacing

dea_limma/workflow/scripts/limma.R

Lines 99 to 103 in 175762a

	# check if the model represented by the design matrix has full rank ie linearly independent columns, which is required to make the model identifiable!
	# new: more efficient and computationally stable compared to the svd() function, especially for large matrices, and it does not require rounding the singular values or checking for non-zero values.
	if(qr(model_matrix)$rank != ncol(model_matrix)){
	stop("Error: The design matrix representing your model does not have full rank, rendering the model not identifiable.")
	}

with limma's function nonEstimable, which shows problematic coloumns and is.fullrank as TRUE/FALSE check.

https://www.rdocumentation.org/packages/limma/versions/3.28.14/topics/is.fullrank