I would like to know if we can use RUVseq for a paired analysis. I have read that the paired analysis need not to be corrected for batches.

so far I am using edger paired analysis:

treat <- as.factor(rep(c("Treat","Untreat"),8))
subjects=factor(c(rep(1:8, each=2)))
design <- model.matrix(~subjects+treat)
y <- calcNormFactors(y)
y <- estimateGLMCommonDisp(y, design,verbose=TRUE)
y <- estimateGLMTagwiseDisp(y, design) 
fit <- glmFit(y, design)
lrt <- glmLRT(fit)

I would like to know if something needs to be done to remove the heterogeneity in the samples as we are dealing with primary human cells.

Including subject in the design, as you have already done in your example code, will control for differences between subjects when testing for differences between treatment, yielding a list of genes that are consistently different between treatments across all subjects. There is no need to use RUVseq for this, unless you suspect that there are some other hidden confounding factors besides subject.in your experiment.