Switching type of cd from unsigned to int make sure the difference
passed to abs() wil be a signed value.

As dc is calculated to at most ((8 * 8 * (2^8-1)) + 32) / 64, ie 255
and the highest value it can hold before it is shifted 6 bit positions
to the right is (8 * 8 * (2^8-1)) + 32) ie. 16352, it should fit nicely
in an integer on any platform where int is at least 16 bit.

Fixes this clang message:

../../../lib/encfrag.c:104:26: warning: taking the absolute value of unsigned type 'unsigned int' has no effect [-Wabsolute-value]
    for(j=0;j<8;j++)sad+=abs(_src[j]-dc);
                         ^
../../../lib/encfrag.c:104:26: note: remove the call to 'abs' since unsigned values cannot be negative
    for(j=0;j<8;j++)sad+=abs(_src[j]-dc);
                         ^~~