Fix sublibrary QA Aqua piracy via treat_as_own for owned types

[ChrisRackauckas-Claude]

Problem

The sublibrary QA groups added in #614 currently fail on master for two of the three sublibraries. The Aqua Piracy test reports the sublibraries' methods as type piracy:

• RecursiveArrayToolsArrayPartitionAnyAll — any/all extended on ArrayPartition
• RecursiveArrayToolsShorthandConstructors — getindex(::Type{VA}, xs...) and getindex(::Type{AP}, xs...)

These are not piracy: ArrayPartition, VA, and AP are types owned by RecursiveArrayTools, which each sublibrary hard-depends on. Extending Base functions on a dependency's owned types is the intended design of these subpackages (it is the entire reason they exist, as documented in their module docstrings).

Fix

Pass the owned types through Aqua's documented treat_as_own mechanism:

Aqua.test_all(RATAPAA; piracies = (; treat_as_own = [RATAPAA.ArrayPartition]))
Aqua.test_all(RATSC;   piracies = (; treat_as_own = [RATSC.VA, RATSC.AP]))

No package source changes; only the QA test configuration is corrected.

Local verification (Julia 1.10, the package's [compat] floor)

Ran the full test/qa/qa.jl (Aqua + JET) for each sublibrary against the in-tree source:

ArrayPartitionAnyAll  QA | Pass 12  Total 12
ShorthandConstructors QA | Pass 12  Total 12

Both groups now pass entirely (Aqua piracy included, JET unaffected).

Out of scope (reported separately)

• RecursiveArrayToolsRaggedArrays QA also fails Aqua piracy (same owned-type pattern) and has genuine JET inference findings (no matching method found similar_type(::Any) in copyto!/fill!/broadcast, because the ragged container's .u field is Any-typed). The JET findings are real and would require type-stabilizing the ragged container, so RaggedArrays is intentionally left out of this focused PR.
• Downgrade (Core) fails with an Unsatisfiable resolver error at the --min=@deps --julia=1.10 downgrade minimum — a compat-floor conflict, separate from this QA issue.

Please ignore until reviewed by @ChrisRackauckas.