Complete python api rewriting

[dos65]

Currently, we can't obtain arguments information from python jobs.
There is initial implementation which possibly can fix that problem(branch)
Also, we need to implement mist-logger for them.
Example:

from mist.decorators import *

@with_args(
    arg('numbers', type_hint=list),
    arg('multiplier', type_hint=int, default=2)
)
@on_spark_context
def simple_context(sc, numbers, multiplier):
    rdd = sc.parallelize(numbers)
    result = rdd.map(lambda s: multiplier * s).collect()
    return {"result": result}

[austinnichols101]

Would like to see a feature to specify the python interpreter as part of the context definition (e.g. PYSPARK_PYTHON, PYSPARK_DRIVER_PYTHON). This avoids having to run one mist instance per python version.

[dos65]

Done.
@austinnichols101 also, nowt there is a way to setup specific python version for context