Switch to 5-year steps by default

[patrickbrown4]

Summary

This tentative PR (to be discussed before merging) proposes a change from 3-year to 5-year steps by default.

Reasons in favor:

• Reduces solve time for the USA_defaults case from 13 hours to ≤8 hours
  • Alternatively, we could instead increase the time resolution (more representative days to help capture near-peak conditions; finer time chunks for better storage representation; perhaps Try applying stress-period PRM via operating reserves instead of load #154).
• We used 5-year steps in NTP; in that case, stakeholders often asked for results for the divisible-by-5 years, so modeling them directly makes that easier
• Minimal impact on the 2035 and 2050 results
• (Also reduces peak GAMS memory use, but probably doesn't affect PRAS.)

Reasons against:

• For scenarios where policies change substantially from year to year, can gloss over shorter-timescale effects
  • But since we don't have inter-year foresight anyway, it's arguable that longer timesteps could avoid some myopic decisions

Technical details

Implementation notes

• I kept the old default (3-year steps from 2020–2050) for the Mid_Case scenario in cases_test.csv so we can make sure it stays working.

Switches added/removed/changed

• yearset: Changed default to 2010..2050..5 (5-year steps)

Validation, testing, and comparison report(s)

Here's a compare report for USA_defaults: results-3-yearsteps,5-yearsteps.pptx

I think the changes to NPV are because we forward-fill non-modeled years instead of linearly interpolating. If we want to regularly compare NPVs for cases with different year steps, we might consider changing that. But by default it shouldn't cause a problem as long as cases with the same year steps are compared (which seems safest to me anyway).

Checklist for author

Details to double-check

• Charge code provided to reviewers
• Included comparison reports for appropriate test cases
• Documentation updated if necessary (double checked and docs do not specify the default year step)

General information to guide review

• Zero impact on results of default case
• No large data file(s) added/modified
• No substantive impact on runtime for full-US reference case (13h -> 8h)
• No substantive impact on folder size for full-US reference case (4.9 GB -> 3.4 GB)
• No change to process flow (runreeds.py, reeds/core/solve/solve.py)
• No change to code organization
• No change to package requirements (environment.yml or Project.toml)

Did you use LLM tools (chatbot or copilot) in the preparation of this PR? If so, describe how

Nope