Aggregate LCOE and NPV over timeslices for objectives

[tsmbland]

Description

Previously, LCOE was calculated at the timeslice-level for the objectives, and then a max operation was taken over the timeslices, and this was the value taken as the cost of installing one unit of capacity.

It seems nonsensical to take the max, and could prevent a technology from being invested in just because it's expensive in a single timeslice.

Really, we should be calculating the average LCOE across the whole year and using this as the cost metric.

To do this, I've added an aggregate_timeslices argument to the LCOE function, which sums timeslice-level costs and production when True, returning a single value for the year (no "timeslice" dimension). This is then what's used for the objectives.

Ultimately, this then gets broadcast over the timeslices here, and the same max operation is taken, which gives you back a single value proportional to the original LCOE for the year as a whole.

It's all a bit convoluted, but this seemed like the easiest way to introduce these changes quickly. I think there are bigger changes required to the solver in the future which will require larger refactoring, so there's no point over-optimizing things right now.

Surprisingly, none of the results change at all. Don't know what to make of this, but it still seems like the right thing to do

Fixes #288

Type of change

• New feature (non-breaking change which adds functionality)
• Optimization (non-breaking, back-end change that speeds up the code)
• Bug fix (non-breaking change which fixes an issue)
• Breaking change (whatever its nature)

Key checklist

• All tests pass: $ python -m pytest
• The documentation builds and looks OK: $ python -m sphinx -b html docs docs/build

Further checks

• Code is commented, particularly in hard-to-understand areas
• Tests added that prove fix is effective or that feature works

[dalonsoa]

Well, it is a bit convoluted, but I don't feel it is a big bottleneck. The key thing is a question of documentation - possibly a separate PR: when LCOE is timeslice-dependent, when it is not and why that is the case.

[tsmbland]

I've added similar changes to NPV/NPC/EAC