Write template metaprogramming

I need a header-only template that computes a compile-time factorial for unsigned integers up to 20 and provides a constexpr function wrapper for runtime use. Make sure the meta-recursive instantiation stops properly and produces a readable static_assert failure when given a negative or too-large value. Include brief usage examples showing compile-time and runtime calls on line numbers matching the declarations.

Before I commit this metaprogram to the library, make it easy for reviewers to accept: surface the maximum supported input (20) at the top, show the compile-time vs runtime API side-by-side, flag where template instantiation could be expensive, and point out the simplest change to generalize to arbitrary integer types.

I wrote a constexpr factorial template and the CI is failing for inputs >20 with a cryptic template error. I need to tell the reviewer what I changed, which tests I ran, and what I fear: that making the bound looser will cause long compile times or overflow. What is the likely diagnosis and the safest next move to fix the failure while keeping compile-time diagnostics clear?

Across several utilities I keep duplicating the same pattern: inline constexpr limits, recursive metafunctions, and ad hoc static_assert messages. This costs review time and causes inconsistent diagnostics. Which habit should I adopt to stop replaying this rework: suggest a single header convention, a compact macro or alias pattern, and one clear rule for test coverage that reduces future churn.