Warnings in IDP

[JoD]

(related to #105 and #153)

It would be good if we had some kind of mechanism to restrict the number of warnings IDP returns to a general user. As it is now, the warnings often confuse users (or at least show up unexpectedly).

E.g. given a vocabulary V with types S en T, one could construct a structure for each interpretation of S and T, and merge both structures before running modelexpand on a theory over V. This seems a perfectly legitimate use of IDP, but it results in two warnings that S and T have empty domains.

Another example are the typederivation warning messages.

Two proposed solutions:

1. a verbosity of warnings: on a low verbosity, only obvious syntax problems (e.g. a double domain element in some interpretation) are printed.

2. output warnings to a separate file, which the user can inspect if he/she suspects something goes awry.