[leburgel]

Following this suggestion: infinite sector types now have the size Unbounded() instead of typemax(Int) in the Manhattan indexing, so that the inclusion-exclusion of #115 leaves out their bounds at compile time. Finite products get faster too; that comes from building the tuple of bounded sizes, but I haven't tracked down why.

	findindex before	after	getindex before	after
U1Irrep ⊠ SU2Irrep	7.6 ns	6.3 ns	190 ns	180 ns
FermionParity ⊠ U1Irrep ⊠ SU2Irrep ⊠ U1Irrep	65 ns	13 ns	830 ns	230 ns
Z4Irrep ⊠ Z4Irrep ⊠ Z4Irrep	29 ns	17 ns	310 ns	220 ns
Z3Irrep ⊠ IsingAnyon ⊠ FibonacciAnyon	31 ns	19 ns	270 ns	205 ns

With SUNRepresentations passing Unbounded() as well (which would replace QuantumKitHub/SUNRepresentations.jl#60), findindex for SU(4) / SU(5) goes from 26 / 65 to 7 / 11 ns, and getindex from 310 / 420 to 175 ns.

Reproducer

using TensorKitSectors, BenchmarkTools
fi(s) = TensorKitSectors.findindex(values(typeof(s)), s)
gi(I, i) = values(I)[i]
for s in (U1Irrep(7) ⊠ SU2Irrep(3), FermionParity(1) ⊠ U1Irrep(-3) ⊠ SU2Irrep(5 // 2) ⊠ U1Irrep(4),
          Z4Irrep(3) ⊠ Z4Irrep(2) ⊠ Z4Irrep(1), Z3Irrep(2) ⊠ IsingAnyon(:σ) ⊠ FibonacciAnyon(:τ))
    @btime fi($s)
    @btime gi($(typeof(s)), $(fi(s)))
end

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/auxiliary.jl	99.07% <100.00%> (+4.12%)	⬆️
src/product.jl	93.60% <100.00%> (-0.89%)	⬇️

🚀 New features to boost your workflow:

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