The ctests are not convenient tests. In this file we can attempt to log the
benefits and costs to using them so we can periodically evaluate which tests we
should keep and which we should get rid of.

Below are some objective costs:

- The time it takes to run the ctests.
- The email discussion back-and-forth.

There are also some non-objective costs. The ctests are annoying and not fun to
maintain, discuss, and fix. Although it's hard to measure this "annoyance"
cost, it should be taken into account when we review.

Below we can try to keep a log of which bugs are found with which ctests. This
way we can see which tests are the most useful. It is difficult to objectively
differentiate between minor bugs and important bugs.

Examples where convergence tests took time and annoyance to figure out that
they were actually "false positives" (i.e., no bug that is worth the time or
complexity to fix):

https://www.mail-archive.com/search?l=mid&q=20210125200655.e27xaaoy2belyxhw%40tallinn


Examples where bugs were fixed (more quickly) because of ctests:

Report : https://www.mail-archive.com/search?l=mid&q=20210126190837.pwuxhrhrikzszfrh%40tallinn
Fix    : 92f8612a

Report : https://www.mail-archive.com/search?l=mid&q=b0a33192bc5635bd1cb5596e346f9c34615ffeb4.camel%40lyx.org
Fix    : ab1d418f

Fix    : ef2b2827

Fix    : 1ddc1c6b

Report : https://www.mail-archive.com/search?l=mid&q=20210218051838.jbwnemfv5ilhlsls%40tallinn
Fix    : 358e4ace

Fix    : 3df42924