I1497 csv io

[ninjapapa]

Fixed #1497

Using the new iomod package to implement csv input and output. Some code was copied from smvinput.py. The original version will be removed in issue #1503.

[laneb]

Do we need a separate connection configuration for the schema? In practice won't the details always be the same as for the CSV file itself?

[ninjapapa]

Schema file will be potential in a place different than csv files. Please see the docstr of _get_schema_connection method. Basically we will allow 4 different ways for user to specify schema of a CSV file:

• In case that the CSV will only be used in a single SMV project, user can code the schema into code through userSchema method
• In case that a CSV might be shared across multiple projects, need to put schema to a file. There are 3 places can store the schema file:
  • If CSV data folder is writable, user can put discovered schema file beside the data file
  • If CSV data folder is not writable, a project level global schema folder can be specified through app config
  • If CSV data's schema is in a specific place which might be different from the global schema folder, user can specify through schemaConnectionName for that special location

This flexibility is needed since some user may want schema to be closer to data, some others may not or may not be allowed.

[laneb]

It's clear from the code and docstring that we're trying to support having the schema file located in a different place than CSV, just not why. That second bullet point is what I was missing. We should add it to the docstring.