`Callback.failed` is always a no-op, and is not called from anywhere.
This PR removes it.

* Add support for separate compilation

* Load Stainless standard library from the classpath

* Fix spurious duplicate loads of Tasty units

* Fix LibrarySuite test

* try to add stainless to path

* new way to add temp folder to options

* opne line

* line break

* print in tests

* change permission of temp folder

* debug

* debug

* trying to reduce path

* temporary fix , with one temp folder per CI, which is cleaned up anyway

* better temp

* add rm folder

* remove print that was causing errors

* add sbt opts

* expand by hand

* Ignore princess solver for LeanMergeSort

* do not run stainless-explicit-fastmachines on PR

* Change polarity of Set to Positive in TypeChecker

* strictly positive

* Add ghost annotation to toList method of Set in library

* change a benchmark to use now ghost toList method

* Bumped compiler version to 3.5.2 and removed a patmat warning

* Depend on Inox module compiled with 3.5.2

* bump sbt plugin

---------

Co-authored-by: Samuel Chassot <samuel.chassot@epfl.ch>

* Some tweaks to propositional logic benchmark

* More tweaks for termination

* update inox
* exclude dependencies to compile
* exclude scala xml 3 and not 2.13

---------

Co-authored-by: Viktor Kunčak <vkuncak@users.noreply.github.com>

Co-authored-by: Viktor Kunčak <vkuncak@users.noreply.github.com>

* fix a bug in the extraction that missed some types hidden behind type synonym definition

* work on the cache using hash

* working on cache using hash

* bug in cache making it using the opposite option (hash vs bin)

* remove debug print

* add flatmap to set

* add flatmap postcondition

* update set

* Move forall and exists to the SetOps class

* move to extension methods instead of implicit class

* new timeout for sat precond test

* new big timeout

* ignore sat-precondition/valid/SATPrecond4 with Princess

This reverts commit 175df63b.