fix: TypeArchive.AddTypes passed the wrong struct to BNAddTypeArchiveTypes

[tinysec]

Problem

TypeArchive.AddTypes marshaled a BNQualifiedNameTypeAndId struct (40 bytes: name, id, type) and passed it to BNAddTypeArchiveTypes, whose contract is BNQualifiedNameAndType (32 bytes: name, type).

The 8-byte layout mismatch made core read the id pointer at offset 24 as the type handle — so a single-element add dereferenced the id string as a BNType* (garbage/crash), and multi-element arrays were fully mis-strided. No type was ever actually added.

core stride (BNQualifiedNameAndType = 32B):  [name 24B][type 8B]
C#    stride (BNQualifiedNameTypeAndId = 40B): [name 24B][id 8B][type 8B]
                                                ^^^^^^^ core reads this as type

The C# P/Invoke comment even said BNQualifiedNameAndType* types — the wrapper just ignored it.

Fix

Match Python TypeArchive.add_type/add_types (typearchive.py:225/222): callers supply only name and definition; the core assigns the id.

• AddTypes(QualifiedNameAndType[]) — builds the correct 32-byte struct.
• AddType(QualifiedName, Type) — convenience (Python add_type).

The id-bearing QualifiedNameTypeAndId struct stays — it is the correct model for the type-reading (BNGetTypeArchiveTypes) and analysis-definition (BNDefineAnalysisTypes) APIs, which legitimately carry ids.

E2E

sharp-binaryninja-tests harness: TypeArchiveAddTypesMarshalingTests — create archive, AddType, assert listable via GetTypeNames; AddTypes second, assert both persist. (Fails on the old 40-byte struct; passes on the fix.)

Verified

• dotnet build -c Release — 0 warnings, 0 errors.
• Harness test green.