That is true.
The intention though is to make it easier for people to translate by handcoding it to C/ C++ or Java by reducing the use of Matlab toolboxes.
However, I agree with you. I will provide 2 versions later on.
The functions would be dramatically faster if you consider the concept of pre-allocation. In modern Matlab releases MLint warns for such problems.
Using BITGET and BITSET is much faster than using DEC2BIN/BIN2DEC and its strange conversion to strings.
Some expressions like "(bitSize*(i-1))+bitSize" could be simplified.