Is it common to find poly(A) tail in RNA-Seq reads? And can this information be used to get better splice variants assignment?

I am not certain of good responses to your two questions, but I have read that length of the poly(A) tail can vary according to conditions within the cell. Thus, noting the length of poly(A), if you have accurate measures from RNA-Seq, between control and various conditions may be an important characteristic of your gene expression data to retain.

I have seen poly A reads in my own limited data.

The following paper has specifically looked at poly A reads as a way to get alternative poly A sites

http://www.nature.com/nature/journal/v464/n7289/full/nature08872.html