[Termtools] Benchmarks for TPDB

[Harald Roman Zankl]

Dear all,

According to the general consensus at WST, TPDB should be divided into
categories each of which is subdivided into benchmarks. Our understanding
of this categorization is as follows:

 - A tool registers for a whole category
   (e.g. TRS, SRS, LP, LP with cuts, FP, ...) and in turn is run on every
   benchmark (e.g. for TRSs: applicative, CSR transformed, imperative,
   (peano) arithmetic, SRS transformed, ...) in this category.

 - To make the competition more thrilling (and shorter) only a (random)
   selection of every benchmark should be used.

As already pointed out by Juergen we also do not want that a tool can
just register for a single benchmark. Nevertheless benchmarking eases
joining the competition since if a tool performs excellent on just a
special class of systems then this will no longer be hidden.

Cheers, TTT2.