Counting number of reads affected by N's in CGIAR

Hi,
I work on plant species. I am using GATK on variant discovery in RNAseq data.
I am not able to decide whether I should use option --filter_reads_with_N_cigar or
-U ALLOW_N_CIGAR_READS.
What do you suggest?

I Would like to count the number of reads affected by N's in the CIGAR? Could you please suggest any tool?

Secondly, I am using Haplotype Caller for variant discovery. It is running very slow (on 12 CPU).
Is it okay to use UnifiedGenotyper instead of Haplotype Caller on RNAseq data?

Thanks