Improve Type System for `pyperf`

Description

• I recently came across pyperf because I want to do some benchmark testing of my changes which I did on an opensource project named ArchUnitPython which are generally to improve the performance of the library. Initially I wanted to start with inbuilt time module but later I found pyperf which kind of can produce good benchmark by handling things like cold start, number of process spawn, warm benchmark etc. But during coding I found that the type system of pyperf is not good as most of most return type and function parameter have Unknown type which kind of hindering my auto completion and understanding of that function or parameter.
• Eg;

• Now it might be issue from pyrefly typechecker but the thing is the library on which I contributed and benchmarking my changes have no Unknown (Till now) same for other libraries like FastAPI, PydanticAI and some more.
• Because of this I am planning to do contribution to the library and improve the type system of pyperf.

If any maintainer could guide me or discuses in this issue, then it would be helpful.

Start by inspecting pyperf's public APIs and the examples shown in the issue, then reproduce the reported Unknown types with a type checker such as pyrefly. Done means the affected public parameters and return values expose useful, accurate types and the reported autocomplete and type-checking problems are resolved.