Python support preparation

[sonalgoyal]

As we are building some functionality in python, need to see what all support we need.

a. zingg.sh update
b. docker update?
c. python dependencies and thier installation and distribution through --py-files?
d. installation documentation update

[sonalgoyal]

Also need to see what all we can suppoort in python. first off, we should expose the model learning metrics as a new phase and use sklearn etc to demonstrate the learning so that users know how to stop. See #254

[sonalgoyal]

We should also support defining the data, output, field definitions and other config params so that people do not have to manually build the config.json.

[navinrathore]

Following are the next set of items planned:

• rename 'client' to 'zingg'
• All the methods of client (Python) should be defined in Java.
  -- BuildAndSetArgument() function should be added
• similar to setData(), new function setOutput() shall be written
• FieldDefinition should take simple quotes as inputs. e.g. "string" rather than ""string"". Internally, needful shall be done
• New MatchType class in python.
• MatchType should be able to passed in FieldDefinition as VarArgs.
• ClientOptions - new class in Python
• new functions in zingg.py - getDFFromDS(), getPandasDFFromDS()
• Arguments python class should have new function writeArgumentsToJson() that will dump Argument object into a json file
• New functions of client.java (*Stat() and getMarkedRecords()) should be added into Python counterpart
• New getUnmarkedRecords()like that of marked Records
• Issue New phase assessModel #296 AssessModel phase should be updated accordingly. All references of '_jvm' should be removed.
• A new phase PeekModel should be introduced. this could be executed to get necessary basic data from Zingg
  -- ZFactory shall have necessary change to return default phase 'PeekModel'