Stale tape detection implementation

[Necried]

This is an implementation for issue #26. While I think the view portion of this code is satisfactory, I did not follow the recommendation mentioned in the issue.

I will expound on the implementation here (this can be copypasted into the issue thread too):

• We still maintain the Array Character representation of tapes. Now consider the two operations
  on individual tapes, Add a character, or Delete a character (from the right):
  1. We do not perform a check on adding. The assumption is that users can only add characters
     defined in the Sigma of the machine. Constant time!
  2. However, when delete is called, we perform the check to see if any offending characters still exist.
     For simplicity, we assume a constant cost of conversion between intermediary data structures.
     The algorithm then becomes: For each character in the array, check if there exists its transition
     in Sigma. Construction of Sigma is linear w.r.t the number of transition edges in the machine.
     The runtime for checking one tape is then O(#transitions + length(tape)).
• We can naively assume that the runtime for checking all tapes (as will happen when onEnter is
  called) will then be O(#tapes *(#transitions + length(tape))).

Storing as CharacterID will save time on checking, but the argument is that #tapes and tape length is small anyway. This part of the code should then be under consideration when we look to optimize.

[CSchank]

I agree we should consider optimizations if they make sense in the future, but given that this is user-defined data, it would be hard for them to define a large enough input to the point where it would make an appreciable difference. Thank you!