"This is why you should write full-coverage testsuites — in the course of testing every path, you are likely to find edge cases that you had not recognized before."

An API (or any code) should have tests written before, during and after completion. Especially after the documentation is completed, to verify.

You're right about the edge cases. They get found in testing, and then become tests themselves.

If one doesn't find issues when running their test suites against their code after significant updates, the test suite is insufficient.