Implement a proper tabular data importer

[alexsb]

When we load tabular data, we should show a proper data importer.

Here's an (old) example of what this could look like:
https://caleydo.org/caleydo-doc/3.1/#!data.md

Things we should be adding:

• A preview of the table
• The ability to skip headers
• The ability to identify the key in the table
• The ability to ignore a column
• Store the data type, among the following types:
  • Numerical
  • Ordinal (ordered categorical)
  • Nominal (unordered categorical)
  • Text/Label
  • Date/Time
  • Set (binary)

We should also ask "is this an edge table" and then give these options:

• Source Node Reference
• Target Node Reference
• Edge Directionality

These datatypes should be auto-deteced, as much as possible, and need to be stored in the database and provided through the API somehow. @kirangadhave has implemented the auto-detection before, we should copy that.