Load CSV file

[rolyp]

We need to support ~20k records for larger examples, perhaps loaded into a 1-dimensional array type, rather than a record, to avoid having to track dependencies on list nodes.

• openDatasetAs' which loads CSV rather than Fluid module
• Migrate existing openDatasetAs to new method

Done/dropped:

• Primitive to open file and return contents as list of records – defer for now as will require interpreter to support I/O
• Use CVS validator to generate a schema, and then use schema to generate a record type?

See also:

• Push ReaderT FileCxt through to foreign op interface #1249

[rolyp]

Might need this if we switch the renewables example to take a dataframe as input.