[JulStraus]

This PR came up as I was a bit uncertain regarding which iteration utilities we included for the different time structures. Most of the work was done by me in #23 but we never really unified it. The work was done by Claude but with extensive review from my side.

Summary

• The iterators returned by strat_periods, repr_periods, opscenarios, and strategic_scenarios are indexable: getindex with a BoundsError check, plus generic firstindex, eachindex, and keys.
  lastindex (and with it x[end] and last) exists only for iterators of strategic and representative periods.
  It is not defined for scenario iterators (scenarios are not ordered) or for the nodes of a TwoLevelTree (a tree has several last nodes).
• All periods, including the Single* periods and StratNode, delegate getindex and eachindex to the structure they contain.
  SimpleTimes and CalendarTimes check the index against len.
• Ordering:
  • StratReprPeriod and StratNodeReprPeriod compare by strategic period, then by representative period.
  • For StratNodes, n1 < n2 holds only if n1 is an ancestor of n2. isless is unchanged, so sort stays reliable and puts ancestors first.
• withprev, withnext, chunk, and chunk_duration forward IteratorSize, so they also work over iterators without a length (e.g. Iterators.filter).
  isfirst is defined for partitions.
  The guards against tree nodes in chunk and chunk_duration also catch calls with keywords.
• Docs: new sections on indexing and ordering in the iteration manual; updated docstrings of TimeStructInnerIter, TimeStructOuterIter, and isfirst.

Changes in behaviour

• <, <=, and >= between StratNodes now consider only ancestors. This fixes start_time and end_time with a TwoLevelTree.
  It also changes the results of downstream code that compares nodes across branches. For example, the CO2Storage capacity constraint in EnergyModelsCO2 with a TwoLevelTree now sums over the ancestors only, which is correct.
  remaining(n, ::TwoLevelTree) for nodes other than the root now sums the node and its descendants. It is still not correct and is left as is.
• sort(repr_periods(::TwoLevel)) now follows the order of iteration; before, it interleaved strategic periods.
• getindex of SimpleTimes and CalendarTimes throws a BoundsError beyond len; before, it returned a period.
• eachindex of all iterators is Base.OneTo(length(x)); before, some methods used the length of internal vectors, which is wrong when len is smaller.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/calendar.jl	94.54% <100.00%> (+3.80%)	⬆️
src/op_scenarios/opscenarios.jl	89.28% <100.00%> (+12.81%)	⬆️
src/op_scenarios/rep_periods.jl	90.19% <ø> (+18.49%)	⬆️
src/op_scenarios/strat_periods.jl	95.65% <ø> (+14.13%)	⬆️
src/representative/rep_periods.jl	90.00% <100.00%> (+2.16%)	⬆️
src/representative/strat_periods.jl	96.07% <100.00%> (+18.71%)	⬆️
src/representative/tree_periods.jl	85.71% <100.00%> (+1.33%)	⬆️
src/simple.jl	100.00% <100.00%> (ø)
src/strat_scenarios/core_types.jl	98.87% <100.00%> (+0.01%)	⬆️
src/strat_scenarios/strat_scenarios.jl	100.00% <100.00%> (+2.63%)	⬆️
... and 4 more

🚀 New features to boost your workflow:

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

[trulsf]

Looks fine. I think the docs update on recommended usage of indexing is a good addition. Even if users are not reading the docs, it may help in the behavior of coding agents.