Also - [ ] - lower RAM usage by fiailed tests descriptions (as we don't see them in UI, only used for ai) - [x] - write failure history in discussion - [x] - set single committer as suspect with 50% confidence - [x] - restore suspect from history - [x] - allow multiple suspects - [ ] do not mark as resolved if there was only problems and next build have failed tests
Also