Fixes

[amirebrahimi]

This PR contains a few small fixes:

• Restrict library sizes to >=256 (due to the constant use of 4 in popcnt and AND_popcnt)
• Don't modify t in next if nCh0.
• For dynamic goal weights in LB and LBB allow enumerating vectors of weight n (handles trivial cases, such as the repetition code generator matrix of a single basis vector of all 1s)
• Fix a bug in SizeRed where the hamming weight was just the weight of E[i].