Make findall faster for AbstractArrays (#37177)

The `findall` fallback is quite slow when predicate is a small function
compared with generating a logical index using `broadcast` and calling `findall`
on it to compute integer indices. The gain is most visible when predicate is true
for a large proportion of entries, but it's there even when all of them are false.
The drawback of this approach is that it requires allocating a vector of `length(a)/8`
bytes whatever the number of returned indices.

Author: nalimilan

base/array.jl

@@ -2142,6 +2142,10 @@ julia> findall(x -> x >= 0, d)
 """
 findall(testf::Function, A) = collect(first(p) for p in pairs(A) if testf(last(p)))
 
+# Broadcasting is much faster for small testf, and computing
+# integer indices from logical index using findall has a negligible cost
+findall(testf::Function, A::AbstractArray) = findall(testf.(A))
+
 """
     findall(A)
 
@@ -2439,7 +2443,8 @@ function findall(pred::Fix2{typeof(in),<:Union{Array{<:Real},Real}}, x::Array{<:
 end
 # issorted fails for some element types so the method above has to be restricted
 # to element with isless/< defined.
-findall(pred::Fix2{typeof(in)}, x::Union{AbstractArray, Tuple}) = _findin(x, pred.x)
+findall(pred::Fix2{typeof(in)}, x::AbstractArray) = _findin(x, pred.x)
+findall(pred::Fix2{typeof(in)}, x::Tuple) = _findin(x, pred.x)
 
 # Copying subregions
 function indcopy(sz::Dims, I::Vector)