Add READMEs for subpackages

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ChrisRackauckas

lib/RecursiveArrayToolsArrayPartitionAnyAll/README.md

@@ -0,0 +1,37 @@
+# RecursiveArrayToolsArrayPartitionAnyAll.jl
+
+Optimized `any` and `all` for
+[RecursiveArrayTools.jl](https://github.com/SciML/RecursiveArrayTools.jl)'s
+`ArrayPartition` type.
+
+`ArrayPartition` stores data as a tuple of arrays. The default `AbstractArray`
+`any`/`all` iterates element-by-element through the partition, which incurs
+per-element partition lookup overhead. This subpackage provides methods that
+iterate partition-by-partition instead, giving ~1.5-1.8x speedup on full scans.
+
+It is separated from the main package because `any(f::Function, ::ArrayPartition)`
+invalidates ~780 compiled specializations of `any(f::Function, ::AbstractArray)`.
+
+## Installation
+
+```julia
+using Pkg
+Pkg.add(url = "https://github.com/SciML/RecursiveArrayTools.jl",
+        subdir = "lib/RecursiveArrayToolsArrayPartitionAnyAll")
+```
+
+## Usage
+
+```julia
+using RecursiveArrayTools
+using RecursiveArrayToolsArrayPartitionAnyAll
+
+ap = ArrayPartition(rand(1000), rand(1000), rand(1000))
+
+# These now use the optimized partition-by-partition iteration
+any(isnan, ap)
+all(x -> x > 0, ap)
+```
+
+Without this package, `any`/`all` use the default `AbstractArray` implementation
+which is correct but ~1.5x slower due to per-element partition indexing overhead.

lib/RecursiveArrayToolsShorthandConstructors/README.md

@@ -0,0 +1,45 @@
+# RecursiveArrayToolsShorthandConstructors.jl
+
+Shorthand constructor syntax for
+[RecursiveArrayTools.jl](https://github.com/SciML/RecursiveArrayTools.jl) types.
+
+This subpackage provides `VA[...]` and `AP[...]` syntax for constructing
+`VectorOfArray` and `ArrayPartition` objects. It is separated from the main
+package because the `getindex(::Type, ...)` method definitions invalidate
+compiled specializations of `Base.getindex(::Type{T}, vals...)`, increasing
+load times for downstream packages that don't need the syntax.
+
+## Installation
+
+```julia
+using Pkg
+Pkg.add(url = "https://github.com/SciML/RecursiveArrayTools.jl",
+        subdir = "lib/RecursiveArrayToolsShorthandConstructors")
+```
+
+## Usage
+
+```julia
+using RecursiveArrayTools
+using RecursiveArrayToolsShorthandConstructors
+
+# VectorOfArray shorthand
+u = VA[[1, 2, 3], [4, 5, 6], [7, 8, 9]]
+
+# ArrayPartition shorthand
+x0, v0, a0 = rand(3, 3), rand(3, 3), rand(3, 3)
+u0 = AP[x0, v0, a0]
+
+# Nesting works too
+nested = VA[
+    fill(1, 2, 3),
+    VA[3ones(3), zeros(3)],
+]
+```
+
+Without this package, use the equivalent explicit constructors:
+
+```julia
+u = VectorOfArray([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
+u0 = ArrayPartition(x0, v0, a0)
+```