Add hybrid report acquisition planner

[masarray]

P2.2 adds an engine-owned RCB capability, partial-coverage, and hybrid report acquisition planner on top of P2.1 main 5b2adcf21a5c82fd45e23672e14ef4edddbd5de2.

Scope

• compose the typed signal catalog with fresh MmsRcbAvailabilityResult evidence;
• calculate static Report Control Block coverage per requested signal rather than treating reporting as an all-or-nothing session decision;
• reuse caller-owned static reports read-only without rewriting DatSet, reservation, or RptEna;
• enable safe configured static BRCB/URCB plans only when the live DataSet binding/directory and explicit free runtime state are proven;
• use verified-empty dynamic BRCB/URCB slots only for still-uncovered signals that resolve exactly in the live MMS directory;
• keep bounded MMS polling as residual fallback only;
• expose typed capability counts, acquisition segments, per-signal assignments, write requirements, warnings, and blockers;
• keep initial fast MMS snapshot acquisition outside the steady-state planner contract.

Safety boundary

• this PR performs no RCB/DataSet writes;
• discovered RCB existence alone never means usable;
• busy, unknown, unreadable, reservation-unknown, and unchecked RCBs are never promoted to automatic write plans;
• dynamic DataSet planning requires a live-verified empty DatSet, RptEna=false, exact availability evidence, and explicit free reservation state (Resv=false for URCB or ResvTms=0 for BRCB);
• static coverage is proven against fresh DataSet member evidence and existing typed catalog membership/reference evidence;
• dynamic member resolution is exact-only (exact MMS reference or one literal live user-reference match); no fuzzy/vendor alias heuristic is used;
• polling fallback and uncovered states never imply signal absence;
• the planner emits typed write intent only; an execution layer must revalidate write-sensitive state before mutation;
• no ARSAS code is changed.

Regression coverage

• mixed static BRCB + static URCB + polling residual;
• caller-owned report reuse without writes;
• static then dynamic URCB coverage with unresolved residual polling;
• dynamic BRCB requiring verified-empty DatSet plus explicit ResvTms=0;
• busy/reservation-unknown RCB rejection;
• effective alternate MMS reference static coverage without changing canonical identity;
• polling-disabled residual remaining Uncovered, never missing;
• capability counts separating discovered, checked, usable, busy, and unknown RCBs;
• public default guardrails and empty-request contract.

Documentation

docs/HYBRID_REPORT_ACQUISITION.md documents the steady-state acquisition sequence, ARIEC/consumer ownership boundary, fresh-availability rules, partial static coverage, dynamic residual planning, polling semantics, write boundary, and physical-validation boundary. CHANGELOG.md records the P2.2 API.

Exact-head validation

Authoritative head: 3407919d80e44b66288fb3c06ee0b6609d3c3499.

.NET CI #301, run 31767731605, passed on this exact head:

• checkout/source provenance, wording and license verification: SUCCESS;
• .NET setup and restore: SUCCESS;
• full Release build: SUCCESS;
• complete automated test step: SUCCESS;
• test diagnostics upload: SUCCESS.

The branch remains exactly based on P2.1 main 5b2adcf21a5c82fd45e23672e14ef4edddbd5de2; main did not advance during this implementation.

Merge / physical validation gate

Keep this PR draft for now. CI proves the deterministic planner and fail-closed software contract; it does not prove that the physical IED accepts the resulting RCB/DataSet operations or actually delivers report traffic. The next integration step is for ARSAS to consume the typed plan without adding app-side report heuristics, then validate on the physical IED that report-covered points transition away from continuous polling while only true residual points remain on MMS polling.

[masarray]

Superseded by current-main consolidation PR #74, merged as 2af88f51df759c7a7083795b3804ab8dc41e8922. The planner, P2.2 tests, and documentation were ported byte-for-byte from this previously CI-green branch onto current main; only CHANGELOG was updated against the current baseline. Fresh PR CI #319 passed, and post-merge main CI #320 passed. Closing this stale draft rather than merging its old-base ancestry. Historical branch/commits remain preserved.