Performance: reimplement bBlock and getBlock in C

[amosr]

The reading and writing accounts for 60% of runtime. Writing these in C should be much faster, and we wouldn't need to spend the extra 16% translating from Haskell to C.

                                                                                                           individual     inherited
COST CENTRE                                       MODULE                                 no.     entries  %time %alloc   %time %alloc

   bBlock                                         Zebra.Serial.Block                    1218          22    0.0    0.0    40.8   36.8
       getBlock                                   Zebra.Serial.Block                     912           0    0.0    0.0    22.4   33.9

     pokeBlock                                    Zebra.Foreign.Block                    988          53    3.0    0.8    16.3   14.1