[FEATURE] extension input validation

[goodwanghan]

Is your feature request related to a problem? Please describe.
NA

Describe the solution you'd like
We need certain level of extension validation, some can be done at workflow compile time and some can be done in runtime, for example

# input has: a,b:str,c,d:int
# prepartition has: a,b
# presort has: c,d desc
# schema: *
def dummy_transformer(df:pd.DataFrame) -> pd.DataFrame:
      return df

Decorator and interfaces should also support this type validation. We should start implementing from the interface level.

input has: can only be verified in runtime, because the schemas can be dynamic. But other validations can be done during the dag construction, it can save significant time to find problems, also it is very expressive.

Describe alternatives you've considered
NA

Additional context
NA

[goodwanghan]

we can use has(contains) and is(equals)