Tabular Data Package -> SQL database should also normalise

[willu47]

Currently, data stored in a tabular data package is not normalised which makes conversion to an SQLite database inefficient.

The values of sets are used as keys to the table, rather than integer keys, which means that it is poor way to store this data, and we obtain few of the benefits of a relational database.

For example, parameter X might look like this:

REGION	VALUE
SIMPLICITY	23.30
REGION 1	154.32
REGION 2	293.2

where the region names are stored in the REGION table:

VALUE
SIMPLICITY
REGION1
REGION2

and a foreign key link is defined in the JSON schema between the parameter X table REGION column and the REGION table.

We would obtain more efficient storage through better use of integer values primary keys in the set tables, and references to these using integer valued foreign keys. For example

parameter X:

fk_REGION	VALUE
1	23.30
2	154.32
3	293.2

REGION table:

primary_key	VALUE
1	SIMPLICITY
2	REGION1
3	REGION2