Unnecessary rate distortion calculations 

```
opj_t1_getnmsedec_sig() method is passed  (bitpos + T1_NMSEDEC_FRACBITS)
and then uses (bitpos - T1_NMSEDEC_FRACBITS) in its computations.

No need to add, and then subtract, T1_NMSEDEC_FRACBITS.  

Likewise for opj_t1_getnmsedec_ref() method.



```

Original issue reported on code.google.com by `boxe...@gmail.com` on 11 Mar 2015 at 12:38
