Improved str.count() algorithm using the Knuth-Morris-Pratt algorithm

[advikkabra]

Implemented the Knuth-Morris-Pratt algorithm for str.count(), similar to the str.find() implementation. This should improve speed significantly in many cases.

[advikkabra]

As an aside, as recommended by #2355, we can improve these algorithms further and benchmark them if need be.
There may be variations of the Boyer-Moore algorithm which work better, as the issue suggests, but I have used KMP as it is previously used for the string find method. I can work on testing and benchmarking them if the current algorithms are to be changed.

CPython implementation
Blog post explaining more

[advikkabra]

@certik Could you review this PR?

[certik]

@advikkabra thank you for this PR. I think it looks great, thanks for implementing the more advanced algorithm, both at compile time and runtime.