Remove SSE4.1 operation, reduce and reorder alignment checks

[ermig1979]

Always use unaligned SSE4.1 loads and stores in binary operations, color and gray reduce, and byte reorder.

• Drop alignment checks and the aligned/unaligned function split in SimdSse41Operation.cpp, SimdSse41Reduce.cpp, SimdSse41ReduceGray2x2.cpp, SimdSse41ReduceGray3x3.cpp, and SimdSse41Reorder.cpp.
• Load and store with _mm_loadu_si128 and _mm_storeu_si128 instead of Load<align>() / Store<align>() and Load<false>() / Store<false>().
• Step-only LoadNose3, LoadBody3, LoadTail3, LoadNose5, LoadBody5, and LoadTail5 are already available in SimdLoadBlock.h and are unused by these functions.