chore: re-stamp calibration fingerprints after the prose alignment

[koen-vg]

#36 merged after the fingerprints in #37 were recorded, so on current main
both tracked artefact sets report [STALE] on all five steps:

[STALE]  feed  (run: tools/calibrate feed)
           workflow/scripts/build_frt_area_attribution.py (code content changed)
           workflow/scripts/build_region_watergap.py (code content changed)
           workflow/scripts/calculate_manure_emissions.py (code content changed)
           ...

I checked every file #36 touched that appears in a calibration DAG. All of them
are comment or docstring edits, except one behaviour-preserving simplification
in get_effective_config (an unreachable pass branch folded into the
condition). The artefacts cannot have moved, so this re-stamps rather than
recalibrates -- tools/calibrate --record for both sets, no solving.

Both sets are green afterwards.

This is the expected cost of hashing code: an inert edit to any script in a
step's DAG marks it stale, and --record is the acknowledgement. Worth knowing
when a docs-only PR touches Python files.