fix(petab): reset a re-injected log observable's noiseDistribution to its v1 base family (#679)

[wshlavacek]

Fixes #679.

Every v1 observable with observableTransformation = log10 converted into a v2 problem the importer then refused, under petab >= 0.9.0:

PybnfError: Observable 'obs_V': observableTransformation 'log10' contradicts the scale of
noiseDistribution 'log-normal'. Give the residual scale in one place -- a log observableTransformation
over a linear noiseDistribution (normal / laplace).

This is the failure behind the three red pytest legs on #678 and on every PR since 2026-09-07. The py3.11 leg stays green only because petab 0.9.0 requires Python >= 3.12.

What was wrong

The scale-preserving converter was leaning on a petab bug. petab1to2 is designed to fold the v1 transformation into the v2 noiseDistribution; a missing return in that merge left the column blank in every petab < 0.9.0, the importer defaulted the blank to normal, and log10 over normal imported as lognormal as ADR-0073 describes. petab 0.9.0 (PEtab-dev/libpetab-python#502) fixed the merge. PEtab v2 has no log10-normal, so the converter now substitutes the natural-log family (log10 + normal -> log-normal, with a warning), a silent ln 10 rescaling of sigma. Stacked under our re-injected log10 column, that is exactly the contradiction the importer exists to refuse.

What changed

petab1to2_preserve_scale now reads each log observable's v1 noiseDistribution (blank -> normal) alongside its transformation, and inject_observable_transformations takes an optional {observableId: 'normal' | 'laplace'} map and resets the row's noiseDistribution to that linear base while writing the transformation. The scale is then stated once, in the preserved column, whichever petab produced the table: a blank cell (0.8.2) and a folded log-normal / log-laplace (0.9.0) both come back to the v1 family. Rows without a transformation keep whatever petab1to2 wrote, so a linear problem is still byte-identical. petab's substitution warning lands inside the catch_warnings block that already silences the parameterScale warning, for the same reason.

ADR-0073 gains an addendum recording this.

Tests

• End to end: a converted log10 observable carries noiseDistribution = normal.
• Injector: log-normal reset to normal with the linear row untouched; log-laplace reset to laplace; the column left alone when no base map is passed.

Three of the four fail against main's convert.py. tests/test_petab_convert.py and tests/test_petab_import.py (175 tests) pass under both petab 0.8.2 and 0.9.0; all tests/test_petab*.py (629 tests) pass under 0.8.2. ruff and the -W docs build are clean.

Not an upstream bug

PEtab v2 removed log10-normal deliberately (PEtab-dev/PEtab#644, libpetab-python#455, editors' decision on PEtab-dev/PEtab#623), and the converter warns about the substitution. The preserved column is the workaround the specification leaves us.