Move trunc out of bond truncation algorithm

[Yue-Zhengyuan]

This PR moves trunc::TruncationStrategy out of bond truncation algorithms ALSTruncation and FullEnvTruncation.

The motivation is to make supporting FixedSpaceTruncation and SiteDependentTruncation in NTU (#144) easier, so I don't need to reconstruct the bond truncation algorithm struct every time I switch the bond to be truncated (since each bond have a different trunc).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/algorithms/truncation/bond_truncation.jl	94.52% <100.00%> (ø)
src/algorithms/truncation/fullenv_truncation.jl	96.77% <100.00%> (ø)

🚀 New features to boost your workflow:

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

[lkdvos]

This does look in general quite different from how we structure the rest of our function signatures, which typically have the truncation strategy in the algorithm... I'm not sure if I really understand the benefit, isn't this just a matter of using @set alg.trunc = ...?

[Yue-Zhengyuan]

Currently the NTU struct looks like

@kwdef struct NeighbourUpdate <: TimeEvolution
    "Bond truncation algorithm after applying time evolution gate"
    opt_alg::Union{ALSTruncation, FullEnvTruncation} =
        ALSTruncation(; trunc = truncerror(; atol = 1.0e-10))
    ...
end

I feel it a bit weird to set trunc = FixedSpaceTruncation() in opt_alg, since the bond_truncate function that ultimately uses opt_alg does not know the original bond space before applying the gate, and FixedSpaceTruncation does not make sense in this narrow context. So I want to make trunc directly a field in NeighbourUpdate.

Though I admit that this is still not ideal, and I'm surely open to suggestions.