Fix padding for multiples of N_BLOCK and pad length calculation

[palto42]

I combined ideas from the pull requests #15 and #16 and updated the examples accordingly. I tested the code on Arduino Leonardo and Raspberry Pi 1B+

Summary of changes:

• The length of the plain data must be calculates without the trailing \0 added by string assignments
  • Note that the function sizeof() includes the \0, so -1 needs to be subtracted for the correct plain data length in case of string input. The function strlen() doesn't count the \0.
  • The "-1" was incorrectly in the function calc_size_n_pad, instead the input length in the example code has been corrected.
• If the length of the plain data is a multiple of N_BLOCK, a full block of padding must be added (instead of none)
• Some small code improvements

[spaniakos]

the changes seem good! i'll give it one two more days to run some test on a crypto-framework i am building and will merge.

thank you.