The reference solution is in the Preloaded instead of Fixture.

Why did you even put his solution as reference and not yours? Your solution is not listed. Do you even have one?

Interesting idea, but I have some remarks:

• random tests are so random, they (almost) never build a valid molecule. Returning false to all random tests would pass 95% of cases :)
• kata is very easy in its current form, it would be much more interesting if it allowed more atoms (and also not single letter ones, like Cl) and variable bonds for, for example, N (3 or 5), S (2 or 4), etc (but then probably random molecules would be 'invalid' from chemical point of view, but IMO meh)