Force specialization on number of args passed to anonymous function constructed by oror and andand

[LilithHafner]

These functions should only accept one argument signature so we should fully specialize.

Fixes #54141

I was also going to make this change for flatten but couldn't demonstrate a performance improvement.

Performance demo (taken from the OP of the linked issue):

julia> versioninfo()
Julia Version 1.12.0-DEV.344
Commit 7ba1b332f87 (2024-04-15 07:40 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (aarch64-linux-gnu)
  CPU: 8 × unknown
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, apple-m2)
Threads: 1 default, 0 interactive, 1 GC (on 8 virtual cores)

julia> using StaticArrays, LinearAlgebra

julia> ab = (a=zeros(SVector{3,Float64}, 1000000),
             b=zeros(SVector{3,Float64}, 1000000));

julia> f_bitwise(ab) = (first.(ab.a).>0) .| (norm.(ab.b) .!= 0)
f_bitwise (generic function with 1 method)

julia> f_boolean(ab) = (first.(ab.a).>0) .|| (norm.(ab.b) .!= 0)
f_boolean (generic function with 1 method)

julia> f_bitwise(ab) == f_boolean(ab)
true

julia> using Chairmarks

julia> @b f_bitwise($ab)
2.282 ms (4 allocs: 122.219 KiB)

julia> @b f_boolean($ab)
51.262 ms (3000004 allocs: 76.413 MiB, 7.88% gc time)

julia> @eval Base.Broadcast function broadcasted(::OrOr, a, bc::Broadcasted)
           bcf = flatten(bc)
           broadcasted(((a, args::Vararg{Any, N}) where {N}) -> a || bcf.f(args...), a, bcf.args...)
       end
broadcasted (generic function with 67 methods)

julia> @b f_bitwise($ab)
2.294 ms (4 allocs: 122.219 KiB)

julia> @b f_boolean($ab)
3.238 ms (4 allocs: 122.219 KiB)