track-bot: test plan

There are several features to test (manually for now... 😱)

1. When an expert first reacts to an issue, the right label should be applied. ✔️
1. Reacting to an issue with some unapproved reaction should have no effect. ✔️
1. Experts can change their mind and the label is appropriately managed ✔️
1. When enough experts agree, the issue should be closed. ✔️
1. When experts disagree, the confusion label should be applied and a comment should be posted which mentions all experts who have reacted to ping them for discussion. ✔️
1. When expert agreement is resolved, the labels should update appropriately. ✔️
1. When the community assessment score exceeds the threshold needed to make an assessment, the appropriate label should be applied. ✔️
1. When the community assessment score falls below the confusion threshold, the confusion label should be applied.
1. When the community assessment changes, the label should be updated appropriately (although confusion may happen first).