function wrappers: remove inferences based on argument names

[aaron-schroeder]

I need to look into this more, but I think I'd prefer a more flexible, less-inferential wrapper to add algorithm functions to the accessor as methods.

Current behavior:

• Algorithm functions that expect series are added as accessor methods by passing them to PositionAccessor._add_series_method.
• This class method creates a method that infers the columns in the DataFrame to use as inputs in the function,
  based on the names of the required positional args in the input function.
• kwargs in the input function are passed as-is.

Desired behavior:

• Functions are replicated as accessor methods as-is , except the positional argument for the DataFrame which becomes implicit.
  Everything else works the same way.
• If specific series are expected by the algorithm functions, don't infer their column names within the DataFrame,
  but provide functionality within _add_series_method to specify the input and output fields.
  More explicit and easier to understand.