[AI-Assisted] feat(dynamics): close built-in capability audit

[EvenSol]

Summary

Advances #3298 WS6 (capability audit closure) under the frozen capability contract.

• classify every built-in ProcessElementInterface declaration of runTransient(double, UUID) by owned state;
• resolve categories from the effective method's declaring class so inherited built-in behavior stays audited while a custom override remains fail-closed;
• add DynamicCapabilityReport.isFullyAudited();
• add a source-inventory CI test that requires every new built-in override to be mapped or cite an existing repository Markdown ADR;
• update the dynamic capability contract and neqsim-dynamic-simulation guidance.

Exact-head and capacity record

• frozen base: master@74c6a3125bdd1f646e7c24921d94b1cc2dfecfb6
• exact head: d6e2285e717c5bc649080119da70c3effdf84c94
• branch: ai/3298-dynamic-capability-audit
• publication occupancy: 4/10 autonomous draft PRs
• active Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 PRs after publication: 1

master advanced after the contract only through the already-audited, non-overlapping refinery PR #3392. This branch deliberately preserves the frozen exact base; it was not rebased or force-updated.

State-ownership audit

• ALGEBRAIC: heater, mixer, splitter, membrane separator, quasi-steady adiabatic pipe, and previously audited algebraic families
• DYNAMIC_LUMPED: filter, committed generator, depressurization, electrolyzer, and previously audited lumped families
• DYNAMIC_DISTRIBUTED: distillation, adsorption/removal beds, pipeline families, and flow-network orchestrators
• BOUNDARY_DYNAMIC: iron-sulfide oxidation source and reservoir boundary state

These labels describe state ownership only. They do not promote numerical, benchmark, safety, controls, line-pack, severe-slugging, or liquid-rich maturity. In particular, PipeBeggsAndBrills remains explicitly unqualified for conservative storage/line-pack claims; those studies remain routed to the qualified TwoFluidPipe path and the unresolved WS1/WS2 gates.

Validation

Passed on the exact submitted tree:

• python devtools/run_spotless.py apply
• python devtools/run_spotless.py check
• bash devtools/run_spotless.sh apply
• bash devtools/run_spotless.sh check
• python devtools/check_documentation_search.py
• focused capability/activation suite:
  DynamicCapabilityBuiltInInventoryTest,DynamicCapabilityReportTest,DynamicCapabilityModuleContainerTest,DynamicActivationReportTest,PressureFlowDynamicCapabilityTest
• Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 regression set:
  ControlsBenchmarkSuiteTest,AgentBenchmarkSuiteTest,SevereSluggingExperimentalBenchmarkTest,SevereSluggingBenchmarkHarnessTest,TwoFluidPipeTransientNullTest,CoupledPressureMomentumTengesdalProgressTest

The local pre-commit executable/module is unavailable, so its configured Spotless and documentation-search hooks were executed directly and passed. An optional full -DskipITs test run progressed without observed failures through the dynamic, distillation, TwoFluidPipe, water-hammer, and terrain/network groups, then was stopped at the campaign fast-validation budget; exact-head draft CI is authoritative for the remaining repository breadth.

Documentation impact

Updated:

• docs/process/dynamic-capability-contract.md
• .github/skills/neqsim-dynamic-simulation/SKILL.md

The docs distinguish full inventory audit from strict runtime preflight and quantitative/public-benchmark qualification.

Frozen stop boundary

No transient equations, solver/IMEX/outlet boundary, regime map, closure, activation policy, benchmark data, or runtime equipment behavior changed. No commercial trace was used. WS1, WS2, WS4, and final routing/qualification work remain open in #3298.

Draft only. Do not mark ready, merge, rebase, force-push, close, replace, or abandon for capacity reasons.