Fix tail extrapolation payloads for conditioned simulation and simulation report

[amirabbas-jalali-bentley]

Description

The conditioned-simulator and simulation-report tasks declare two fields in their distribution parameter as required but nullable, with no default:

• distribution.tail_extrapolation
• distribution.tail_extrapolation.lower

So the key has to be present, but null is a valid value for it. TaskRunner serialises parameters with exclude_none=True, so whenever either field was left unset the key was dropped and the task rejected the job:

• a distribution with weights but no tail extrapolation failed with 'tail_extrapolation' is a required property
• a tail extrapolation with only an upper tail failed with 'lower' is a required property

There was no way around it from the calling side, since passing None explicitly makes no difference to exclude_none.

Changes

• DistributionParams, SimulationReportDistribution and TailExtrapolationParams now always emit these keys, as null when unset. This is the same @model_serializer(mode="wrap") approach DeclusteringParameters already uses to keep power: null. Every other field is still omitted when it's None, and no other task is affected.
• SimulationReportDistribution.tail_extrapolation was annotated as None, so a real tail extrapolation couldn't be passed to the simulation report at all. It now takes the same TailExtrapolationParams as the conditioned simulator, since both tasks accept the same shape.

I considered switching TaskRunner to exclude_unset=True instead, but it doesn't help on its own (a field the caller never set is still dropped), and it would stop sending the client-side defaults several tasks rely on, such as the kriging method and the conditioned simulator's defaults.

No breaking changes. Payloads that were already accepted are unchanged; only the ones that used to be rejected now carry the extra null keys. lower stays optional.

Testing

• New tests pin the explicit nulls on the wire for both tasks, check that a real lower isn't overwritten, and check that the simulation report takes a real tail extrapolation. Four of the five fail without the fix.
• Validated the resulting payloads against both tasks' JSON schemas: no distribution, a distribution without tail extrapolation, an upper tail only, and both tails are all accepted.
• evo-compute tests: 418 passed. ruff check and ruff format --check are clean.

Checklist

• I have read the contributing guide and the code of conduct

[Copilot]

Copilot review overview

🟢 Approval recommended

The reviewed changes have regression coverage and no unresolved issues.

Review effort: Balanced
Findings: None

What changed in this PR

Fixes serialization of required nullable tail-extrapolation fields for conditioned simulation and simulation-report payloads.

Changes:

• Preserves required null keys during serialization.
• Enables tail-extrapolation parameters in simulation reports.
• Adds regression tests and updates generated artifacts.

File	Description
uv.lock	Synchronizes the evo-objects workspace version.
packages/​evo-compute/​tests/​geostatistics/​test_simulation_report_tasks.py	Tests simulation-report tail serialization.
packages/​evo-compute/​tests/​geostatistics/​test_conditioned_simulator_tasks.py	Tests nullable and populated tail payloads.
packages/​evo-compute/​src/​evo/​compute/​tasks/​geostatistics/​simulation_report.py	Adds tail-extrapolation support and null-preserving serialization.
packages/​evo-compute/​src/​evo/​compute/​tasks/​geostatistics/​conditioned_simulator.py	Preserves required nullable distribution fields.
mkdocs/​site/​packages/​evo-compute/​typed-objects/​simulation-report/​SimulationReportDistribution.html	Updates generated API documentation.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[GriffinBaxterSeequent]

All good from a evo-python-sdk-maintainers perspective!