let pl.Expr.reshape optionally take a tuple or 2 scalars rather than just a tuple

[deanm0000]

Description

Most methods in polars (with_columns, select, group_by, etc) take *args and are flexible to using that or inputting a Sequence. reshape is the outlier in that it specifically requires a tuple. Most times, for me anyway, I just want the rows to be inferred so it'd be nicer to do pl.col('a').reshape(width) rather than pl.col('a').reshape((width, -1)). In comparison to numpy, its reshape accepts an int as the shape not just a tuple.

I'd like to redefine the function to

from polars.exceptions import InvalidOperationError
def reshape(self, dimensions: tuple[int, int] | int, *args: tuple[int, ...]) -> Expr:
    if isinstance(dimensions, int):
        if len(args)==0:
            dimensions=(dimensions, -1)
        elif len(args)==1:
            dimensions=(dimensions, args[0])
        else:
            msg="reshape only takes 2 dimensions"
            raise InvalidOperationError(msg)
    else:
        if len(dimensions)!=2:
            msg="When dimensions is tuple, it must be 2 elements wide"
            raise InvalidOperationError(msg)
        if len(args)>0:
            msg="When dimensions is tuple, extra arguments are ignored"
            issue_warning(msg,...)
        
    return self._from_pyexpr(self._pyexpr.reshape(dimensions))