Add RLE fast path for single-byte runs (74.1% faster)

Add RLE fast path for single-byte runs (74.1% faster) #68