Annotated Corpus Map: add heuristic for choosing proper t-SNE attributes

[ajdapretnar]

Is your feature request related to a problem? Please describe.
Annotated Corpus Map does not identify t-SNE attributes automatically. It takes the first two meta attributes instead, which is wrong when there are previous meta attributes.

Describe the solution you'd like
ACM automatically identifies t-SNE attributes if they exist in the domain.

Describe alternatives you've considered
/

[ajdapretnar]

Fixed in #1046.