[JortGroen]

What

Four lines in the tutorial helpers, so they work with both pandas 1.5 and pandas 3, plus one line in the CI environment:

• tutorials/power_balance/functions_T1.py (2×) and tutorials/physical_congestion/functions_T3.py (1×):
  result_pd_df.index = pd.to_datetime(result_pd_df.index) → result_pd_df.index = result_pd_df.index.map(pd.Timestamp)
• tutorials/power_balance/functions_T1.py: resample('H') → resample('h')
• environment.yml: add nest-asyncio. Separate commit, unrelated to pandas: ipykernel 7 now depends on nest-asyncio2 instead of nest_asyncio, so in a fresh environment the notebooks' import nest_asyncio failed and every test errored. dev has the same problem; its last green run was in June, before this change.

Why

• Illuminator's Collector writes the first output row as a bare date (2012-06-01) and later rows with a time (2012-06-01 00:15:00). pandas ≥ 2 infers one format from the first value, so pd.to_datetime fails with ValueError: unconverted data remains ... " 00:15:00". index.map(pd.Timestamp) parses each value on its own.
• pandas 3 removed the 'H' frequency alias.
• Illuminator's check-example-compatibility.yaml runs this repo's dev tests against Illuminator PR branches. The Illuminator Python 3.14 + pandas 3 upgrade (issue 259, branch 259) can't pass that check until this is in dev.

pandas 1.5 / 3 compatibility

• format="mixed" (the fix used in Update Python version from 3.11 to 3.14 #25) doesn't exist in pandas 1.5.3, which the PyPI release of illuminator (3.1b0) pins and this repo's CI installs. index.map(pd.Timestamp) works in both and returns a DatetimeIndex.
• 'h' is accepted by pandas 1.5 as well.
• Tested locally: full pytest -v suite passes on Python 3.12 + pandas 3.1.1 + Illuminator branch 259. This PR's CI covers pandas 1.5 (PyPI illuminator).

Relation to #25

Supersedes only the helper part of the closed #25. That PR also moved the env to Python 3.14 + pandas 3 + illuminator @ git…@259, which is left out here (see follow-ups).

Follow-ups (not in this PR)

• Remove nest_asyncio (import nest_asyncio / nest_asyncio.apply()) from the notebooks, together with the Python 3.14 / pandas 3 env move from Update Python version from 3.11 to 3.14 #25, after an Illuminator release that includes the new Simulation.run() fix (runs mosaik in a worker thread when Jupyter already has an event loop). The published illuminator still needs nest_asyncio, so it stays for now. On Python 3.14, nest_asyncio gives silently wrong simulation results. Note that the in-progress branches tutorial-5, tutorial-5-ide-integration and ide-case-final add more notebooks that use it.
• tutorials/residential/Illuminator_BSc_Tutorial.ipynb (~line 443): pd.to_datetime(df_latest5['date'], errors='coerce') has the same mixed-format problem. Under pandas 3 every row after the first becomes NaT (checked: ['2012-06-01', '2012-06-01 00:15:00'] → [Timestamp('2012-06-01'), NaT]). Not covered by the test suite.

🤖 Generated with Claude Code