in RUVseq manual they state for the downstream analysis with edgeR to use following design matrix:
design <- model.matrix( ̃x + W_1, data=pData(set1)) , where W_1 are the coefficients of unwanted variance

In edgeR manual batch effect should be the first in the design matrix:
design <- model.matrix(~Batch+Treatment)

Also, when I perform DESeq2 analysis as with the
DESeqDataSetFromMatrix(countData=signal, colData=Design, design=~condition+batch)

I get following header from the results :
log2 fold change (MAP): batch 
Wald test p-value: batch 

If I use instead DESeqDataSetFromMatrix(countData=signal, colData=Design, design=~batch +condition)
I get following and it seems to be appropriate:
log2 fold change (MAP): condition WEN vs WNN
Wald test p-value: condition WEN vs WNN

What would be a correct way to include factors of unwanted variance in the deseq and edger design matrix?

In the DESeq2 vignette, we recommend putting 'condition' at the end of the design, so you can use results() without extra arguments (although you can extract any results you like using the arguments)

I'm not sure why it says "condition batch" as you haven't provided the code for how you generated this results table. What code did you use (also sessionInfo)?