Reduce invalidations of other methods

[timholy]

These changes, combined with more extensive changes to Julia itself, greatly reduce latency stemming from loading FixedPointNumbers. Ref JuliaLang/julia#35733.

There will be very little benefit to this on its own, but we can at least find out if it works across Julia versions.

[kimikage]

julia> sizeof(Fixed{BigInt,10})
8

😝
This is off-topic but it doesn't seem to be well documented what sizeof should return. (https://docs.julialang.org/en/v1.4/base/base/#Base.sizeof-Tuple{Type})
I saw some troubles in the Discourse.

[kimikage]

FixedPointNumbers.jl/src/FixedPointNumbers.jl

Line 6 in f0f9e74

zero, oneunit, one, typemin, typemax, floatmin, floatmax, eps, sizeof, reinterpret,

We will not need sizeof, anymore.

[timholy]

Oof. That's not good. I'm surprised that Core.sizeof does that.

NVM. The result is pretty much undefined when it's not inline storage. In a sense the new answer is better.