Avoid string StartsWith and EndsWith

[tsholmes]

String functions in mono are extremely slow. This avoids calling StartsWith and EndsWith for cases where we could just check the characters directly.

This had a varying effect from almost no speedup to an 8% speedup in my test cases depending on how call heavy the script was (the EndsWith("()") check is the most expensive).

[Dunbaratu]

Let me know what you think, @tsholmes. I am ready to merge this, but was wondering about this one point.

[Dunbaratu]

Should we just have a generic "endswith" and "startswith" call in our utility library that operates on generic strings but does it char by char? Would it be an improvement over the generic mono one that way? i.e:

public static bool EndsWith(string biggerString, string substring)
{
    int len = subString.Length;
    int i = biggerString.Length - len;
    if (i<0) return false;
    while (i < len)
    {
        if (subSrring[i] != biggerString[j]) // maybe allow for culture case-insensitive comparison here?  Nah, probably not.
            return false;
        ++i;
        ++j;
    }
    return true;
}

It seems weird to be writing one hardcoded primitive method per endswith comparison we want to do, but I can conceed it may be worth it if it's faster that the generic solution.

[Dunbaratu]

I'm going to switch to counting this forward to save some math. (set index to strLen - i - 1 at first, then just increment it by +1 each iteration).