Add SmallTag type for more compact Dual types

[topolarity]

This is largely to improve printing and reduce visual noise for packages with large function types (e.g. SciML)

Compare tag = :small:

ForwardDiff.Dual{ForwardDiff.SmallTag{0x39f35d61d979c3d1}, Float64, 3}

to tag = :default:

ForwardDiff.Dual{ForwardDiff.Tag{var"#f#1"{DAEProblem{Vector{Float64}, Vector{Float64}, Tuple{Float64, Float64}, false, Vector{Float64}, DAEFunction{false, SciMLBase.FullSpecialize, typeof(f), Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing, Nothing, Nothing}, @Kwargs{}, Vector{Bool}}, BrownFullBasicInit{Float64, Nothing}, DFBDF{5, 0, ADTypes.AutoForwardDiff{nothing, Nothing}, Nothing, OrdinaryDiffEqNonlinearSolve.NLNewton{Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}}, typeof(OrdinaryDiffEqCore.DEFAULT_PRECS), Val{:forward}(), true, nothing, Nothing, Nothing}}, Float64}, Float64, 3}

[codecov]

Codecov Report

Attention: Patch coverage is 86.36364% with 9 lines in your changes missing coverage. Please review.

Project coverage is 86.92%. Comparing base (c310fb5) to head (d88bf58).
Report is 21 commits behind head on master.

Files with missing lines	Patch %	Lines
src/config.jl	86.36%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #748      +/-   ##
==========================================
- Coverage   89.57%   86.92%   -2.66%     
==========================================
  Files          11       10       -1     
  Lines         969     1025      +56     
==========================================
+ Hits          868      891      +23     
- Misses        101      134      +33     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[mcabbott]

We could also revive #572, closed it was thought that SciML printing problems were solved in Base.

Can you comment on the comparison?

[topolarity]

We could also revive #572, closed it was thought that SciML printing problems were solved in Base.

Base has type-folding now that helps many stack traces, but you still sometimes need a repr(...) of a type that you can copy-and-paste, which disables folding

Can you comment on the comparison?

The main difference is that debugging perturbation issues without the actual function / array types can be very confusing, so it seems useful to keep the original functionality around. Other than that, the implementation should be mostly equivalent (the hash is based in many cases on the objectid)

Also it's worth mentioning that tag = short does not allow the Dual / Config type to be inferred on 1.10 sadly, since hash(::Type) is only const-eval-able on 1.11+ (see JuliaLang/julia#52427 (comment))