There are certain sweet spots, but this should not be overemphasised.
Square numbers of WH strength are nice, n^2+n is the next best thing. There's no reason to stick to this slavishly, though.
If we have other tight design constraints, fitting a sweet-number warhead may not be practical, (e.g. "slow but high-yield size-1 ASM", "the smallest missile that can fit a practical sensor to avoid overkill")
Serious optimisation forcing you to make samey ships in all campaigns is a red herring.
Most of the numeric optimisation here is about good design practice for specific implementations, it doesn't restrict our high-level goals.
The theory behind propulsion optimisation is the same whether you're designing a missile, a beam cruiser or a freighter... and many people make the same blunders for all of them. This doesn't make things richer or deeper, because all their clever little decisions are rendered meaningless - their creation does its job poorly or much more expensively than necessary.
However, if you are aware of the reasonable trade-offs and juggle compactness, build cost, tech requirements, fuel use, redundancy, projected future use when the design is no longer fresh... the decisions become more meaningful, there are some real trade-offs rather than just a scale of incompetence.
Observing good practice for the fiddly details doesn't mean your concept needs to be samey, you can still implement any doctrine you care to envision. Your vessels will just share some general characteristics, just like many vehicles share circular wheels - they simply work better than triangular ones for most common needs.
The cool bit: If you find a good reason to break multiple rules of usual good practice, and your oddball design has notable advantages over the "strictly formula" approach, you have some true innocation rather than yet another fashion folly.