Write tests to ensure `LongTable` -> `data.table` -> `LongTable` is symmetric (i.e., doesn't lose data)

[ChristopherEeles]

In manual testing I discovered that this symmetry is satisfied for the real NCI-ALMANAC LongTable object, but no the merckLongTable example data object.

Need to confirm that this is an issue with the example data object and not an issue with the coercion methods.

Also noticed that:

• metadata needs to be added to data.table, or metadataMap has to set to empty in as(LongTable, 'data.table')
• need to delete the longTableDataMapper attribute from a data.table during coercion back to LongTable or it ends up as an attribute of every data.table slot in the LongTable

[ChristopherEeles]

Symmetry is true with MIT-MELANOMA.