[feature] Implement tidyselect

[talegari]

Column names should also support functions along with strings:
df.select(['a', 'b', 'c']) # regular
df.select(['a', 'b', starts_with('c'), ends_with('d'), contains('some_regex')])

tidyselect should power all the methods that take column_names as the input.

[grahitr]

Also, replace_na, should be changed to accept list of columns and/or column selector in the key of the dictionary being passed.

Also, is the value appropriate argument name in replace_na?

[talegari]

IMHO, value should be a single value for simplicity. If we allow a list of values to be passed, then we implicitly already the columns to be renamed right?

my suggestion:
df.replace_na({ends_with("width"): 0}) and not df.replace_na({ends_with("width"): [0, 1]})
as in latter case, usually we might not know how many columns get selected.

[grahitr]

To expand tidyselect for following methods

count
add_count
nest_by
expand
complete
Summarise -> in by
Mutate -> in by

[sardnar]

Is there is an easy way to filter using tidyselect. Like:

df.filter(starts_with("x") < 10)