The `subset` relation contains regions subsets of themselves

[lqd]

We might need to investigate the symmetric subset tuples produced, if we can remove them, and if this has performance implications: in the clap benchmark, we can see 480K regions subsets of themselves.

To see these tuples, we can get debug output like the following (for example, right before filling the output tuples in -v / dump_enabled mode):

println!("subset ({}) symmetries: {:?}", subset.len(), subset.iter().filter(|&(r1, r2, _)| r1 == r2).count());

Outputting these results over the clap dataset:

1. subset (7531526) symmetries: 794833 for the Naive variant
2. subset (2714121) symmetries: 481294 for the DatafrogOpt variant