[stdlib] Extract from / Load to files standard types

[hartym]

Although using pandas function names was the first idea, I'm not so much inclined to use lowercase names anymore because we have to implement those as classes anyway. So there are two options, either we drop pandas names completely (which sounds not so unacceptable), or make them available through the bonobo.compat.pandas module, which will just alias things.

Extract from file ...

• [x] FileReader (default: one line is one row)
• [x] CsvReader (alias: read_csv)
• [ ] ExcelReader (alias: read_excel)
• [x] JsonReader (alias: read_json)
• [ ] HtmlReader (alias: read_html)
• [ ] ClipboardReader (alias: read_clipboard)
• [ ] PickleReader (alias: read_pickle)

Load to file ...

• [x] FileWriter (one row, one output line)
• [x] CsvWriter (alias: to_csv)
• [ ] ExcelWriter (alias: to_excel)
• [x] JsonWriter (alias: to_json)
• [ ] HtmlWriter (alias: to_html)
• [ ] ClipboardWriter (alias: to_clipboard)
• [ ] PickleWriter (alias: to_pickle)

Still need to stabilize / freeze this API.

[hartym]

Retained option will be to drop pandas names completely, although a bonobo-pandas or somehow similar package may exist in the future to re-create the aliases. A documentation page should be added documenting bonobo "for pandas users".