[amirabbas-jalali-bentley]

Description

#354 made the conditioned-simulator and simulation-report payloads always include two keys, sending null when they were unset:

• distribution.tail_extrapolation
• distribution.tail_extrapolation.lower

Both tasks used to require these keys to be present. They now treat both as optional, defaulting to null, so the workaround is no longer needed. This PR removes it, so unset values are left out of the request like every other optional field.

Changes

• Remove the @model_serializer(mode="wrap") overrides from TailExtrapolationParams, DistributionParams and SimulationReportDistribution, along with the imports only they used.
• Update the payload tests to expect the keys to be left out. Two of them are renamed from ..._sends_null to ..._omits_key.

SimulationReportDistribution.tail_extrapolation keeps the TailExtrapolationParams | None type from #354, so the simulation report still accepts a real tail extrapolation.

No breaking changes. Models, fields and defaults are unchanged. The only difference on the wire is that an unset tail_extrapolation or lower is now left out instead of being sent as null.

Testing

• Validated the resulting payloads against both tasks' current parameter schemas. All four cases are accepted: no distribution, weights without tail extrapolation, an upper tail only, and both tails.
• Ran a conditioned simulation followed by a simulation report end to end, once with no tail extrapolation and once with only an upper tail. All four jobs completed.
• evo-compute tests: 517 passed. ruff check and ruff format --check are clean.

Checklist

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