TTD driven version of stack-overflow

[Konard]

Instead of question - person writes an automated test that checks that solution works.
Multiple possible solutions that can pass the test are added and automatically tested.
All input is well-come does not matter from AI or human. Best solutions are voted up.
Solutions can be sorted by memory usage, performance, and language may be chosen.
All tests are language agnostic, so the test works using CLI input/output.
The test can come with a description, that is not fully represented in the test.
Each question is freezed when first solution passes the test.
If user apply with similar but different test suite or description that changes meaning or logic it should be not considered as duplicate, but anyway it can linked to similar tests with NarrowerThan and BroaderThan links.