fix(DC0007,DC0008,DC0004): document interfaces and control add-ins

[MODUSCarstenScholling]

Summary

Fixes #451

• Restore DC0004 diagnostics for undocumented control add-in procedures.
• Report undocumented control add-in event declarations as DC0009 with event-specific wording.
• Remove the unavailable local/internal remediation from the public-procedure diagnostic.
• Extend DC0007 and DC0008 object analysis to interfaces and control add-ins by analyzing IObjectTypeSymbol.
• Preserve the test-codeunit exemption by applying it only to symbols that also implement IApplicationObjectTypeSymbol.
• Report DC0007 for undocumented public interfaces and control add-ins, and DC0008 for undocumented internal interfaces.
• Add regression coverage for documented and undocumented interface/control-add-in objects, plus parameterized control-add-in events.
• Remove the vacuous internal-procedure control-add-in fixture.
• Document the corrected control-add-in procedure/event diagnostic routing in the analyzer and documentation repositories.

[Arthurvdv]

Review of #507 — document interfaces and control add-ins

Solid, well-scoped fix; the object-level IObjectTypeSymbol widening and the separate EventDeclaration action are the right shapes, and the fixtures/docs are updated. A few things worth addressing before merge:

1. DC0004 message is inaccurate for control add-in events (and procedures). AnalyzeControlAddInEvents reports PublicProcedureRequiresDocumentation, so an undocumented event ControlAddInReady() surfaces as:

Public procedure 'ControlAddInReady()' must include XML documentation comments (or be restricted to local or internal scope).

It is not a procedure, and control add-in members take no local/internal modifier, so the remediation is impossible. DC0009 (EventRequiresDocumentation, "Event '{0}' ...") is enabled by default and reads correctly for events. Worth reconsidering the ID for events, or at least a control-add-in-appropriate message. (The same "restricted to local or internal scope" tail is now emitted for control add-in procedures too.)

2. Missing NoDiagnostic regression for documented interfaces/control add-ins (object rule). ObjectRequiresDocumentation gained PublicHasDiagnostic/InternalHasDiagnostic fixtures only. There is no fixture proving a documented public/internal interface or control add-in is left alone, so a future false positive on these IObjectTypeSymbol-only kinds would go uncaught.

3. The parameterized-event path is untested. Every control add-in event fixture is parameterless, so GetEventDisplayText/GetTypeDisplayText — including the #if NETSTANDARD2_1 reflection branch — never execute under test. A HasDiagnostic fixture with event Foo(bar: Integer) would cover it.

4. InternalNoDiagnostic/ControlAddIn.al now passes vacuously. It still holds an undocumented procedure MyProcedure() and asserts no DC0006. With the exemption removed, that procedure now emits DC0004, not DC0006 — so the test stays green without validating anything meaningful. Consider updating or removing it to reflect the new behavior.

5. Minor: display-text duplication. GetEventDisplayText/GetTypeDisplayText re-implement what the method path already gets from GetDiagnosticDisplayText(MethodSymbolDisplayFormat.MethodSignature); the two formatters can drift over time.

Nothing here is a crash; items 1–2 are the ones I'd want resolved before merge.