Allow UDFs to take all columns in `reduce()`

[hombit]

Feature request

Currently, we require the user to specify all reduce() input columns explicitly, for the sake of performance. Let's allow a special case when no columns are specified and all items per row are passed as a single dictionary:

def udf1(base_col, sub_col):
    # base_col is a single value, sub_col is a numpy array
    ...

# Existing behavior:
nf.reduce(udf1, "ra", "lightcurve.time")

def udf2(row):
    assert isinstance(row, dict)
    assert "ra" in row and "lightcurve.time" in row
    # row is a dictionary: base columns are single values,
    # nested columns are numpy arrays
    ...

# New behavior:
nf.reduce(udf2)

Before submitting
Please check the following:

• I have described the purpose of the suggested change, specifying what I need the enhancement to accomplish, i.e. what problem it solves.
• I have included any relevant links, screenshots, environment information, and data relevant to implementing the requested feature, as well as pseudocode for how I want to access the new functionality.
• If I have ideas for how the new feature could be implemented, I have provided explanations and/or pseudocode and/or task lists for the steps.

[hombit]

We may also change the current interface so the default udf input is a dict

def map_rows(
    self,
    *,
    # list of the columns to pass, None - all, str - single , list[str] - few
    args: None | str | list[str],
    # how udf will be called: dict - udf({"col1": value, ...}, **kwargs), args - udf(value, ..., **kwargs)
    row_container: Literal['dict'] | Literal['args'],
    # None - output should be a dict (or 0, 1, 2 will be used for columns), str or list[str] - output is single value or a tuple, names are specified by this arg
    output_names: None | str | list[str],
    append_columns: bool = False,
    # Respects output_names
    infer_nesting: bool = True,
    **kwargs,
)

[dougbrn]

Working on above, have noticed that args might not be the best argument name, as it's conventionalized to mean something specific within a python function itself. Certain weirdness occurs when trying to inspect "args" within a debugger for example. Given that args in our case only refers to args being supplied by columns from the nestedframe, I currently have the name changed to "columns". You could argue for other names like "func_args", but technically a function may have arguments that are not supplied from the nestedframe as well, so "columns" seems a bit better. Maybe there's a better name than "columns" though!

[hombit]

@dougbrn I agree about columns, and we also may keep the old *columns behavior if you prefer that

[dougbrn]

So far, keeping columns as a contained list of column names is looking cleaner than allowing args to be passed independently imo #366

[hombit]

@dougbrn ok with me!