Add randomly generated tests

[drinckes]

This adds 1000 random tests with random amounts of decimal places and code lengths.

[fulldecent]

Can you please verify one of the code length 15 examples by hand. Using a calculator with sufficient digit precision.

[slpat]

We appreciate everything that you do.

[drinckes]

I think I'm going to pause this for the moment. A higher priority is getting the CI tests working and then I may add more test cases.

[fulldecent]

@drinckes got it can you confirm that the testing on C & CPP is the right approach now?

#599

Or are further changes needed on those two implementations before we fix the other ones

[drinckes]

Lots of issues in the tests generated. There's lots of boundary points, so we're getting variations in codes due to floating point handling in different languages (even on the same hardware).

The existing Java implementation seems to think that the minimum valid code length is 4 (spec says 2) (#667).

I'm going to generate a set that encode, then we'll see what fails on e.g. M1 mac I guess.

[drinckes]

NB: Blocked by #667

[fulldecent]

Good notes

Actually, only two floating point operation are required. It's a modulation (if needed) then multiplication.

This approach is spelled out in your favorite PR.

I could try this in the C implementation. Just because I'm guessing that is the canonical one since the test runs first.

But if I'm going to work on an implementation, to be honest, I would want to get my hands dirty and do a rewrite or fix a few things while I'm working on it.