optimizer: fix #42754, inline union-split const-prop'ed sources

[aviatesk]

This commit complements #39754 and #39305: implements a logic to use
constant-prop'ed results for inlining at union-split callsite.
Currently it works only for cases when constant-prop' succeeded for all
(union-split) signatures.

example

julia> mutable struct X
           # NOTE in order to confuse `fieldtype_tfunc`, we need to have at least two fields with different types
           a::Union{Nothing, Int}
           b::Symbol
       end;

julia> code_typed((X, Union{Nothing,Int})) do x, a
           # this `setproperty` call would be union-split and constant-prop will happen for
           # each signature: inlining would fail if we don't use constant-prop'ed source
           # since the approximated inlining cost of `convert(fieldtype(X, sym), a)` would
           # end up very high if we don't propagate `sym::Const(:a)`
           x.a = a
           x
       end |> only |> first

before this commit

CodeInfo(
1 ─ %1 = Base.setproperty!::typeof(setproperty!)
│   %2 = (isa)(a, Nothing)::Bool
└──      goto #3 if not %2
2 ─ %4 = π (a, Nothing)
│        invoke %1(_2::X, :a::Symbol, %4::Nothing)::Any
└──      goto #6
3 ─ %7 = (isa)(a, Int64)::Bool
└──      goto #5 if not %7
4 ─ %9 = π (a, Int64)
│        invoke %1(_2::X, :a::Symbol, %9::Int64)::Any
└──      goto #6
5 ─      Core.throw(ErrorException("fatal error in type inference (type bound)"))::Union{}
└──      unreachable
6 ┄      return x
)

after this commit

CodeInfo(
1 ─ %1 = (isa)(a, Nothing)::Bool
└──      goto #3 if not %1
2 ─      Base.setfield!(x, :a, nothing)::Nothing
└──      goto #6
3 ─ %5 = (isa)(a, Int64)::Bool
└──      goto #5 if not %5
4 ─ %7 = π (a, Int64)
│        Base.setfield!(x, :a, %7)::Int64
└──      goto #6
5 ─      Core.throw(ErrorException("fatal error in type inference (type bound)"))::Union{}
└──      unreachable
6 ┄      return x
)

---

This PR also comes with 774ca28, which
improves code qualities of inlining code, most notably eliminates
excessive specializations around Pair{Any, Any} construction/zipping.

[aviatesk]

IIRC this change will require a manual cherry pick for backporting.

@nanosoldier runbenchmarks("foldl" || "sort" || "union", vs = ":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here.

[KristofferC]

Would be good to get it backported to prevent the performance regression.

[aviatesk]

ok I'll take it once we merge this PR.

[aviatesk]

I also added general improvements around the inlining code as well.
Especially, 774ca28 eliminates excessive specializations around Pair{Any, Any} constructions and zipping Vector{Pair{Any,Any}}.
As a result, the # of precompiled statements is reduced to 1469/1504 from 1483/1518.

[vtjnash]

SGTM

[aviatesk]

I found we need to make sure we don't inline !isdispatchtuple(sig) case when union-splitting.
MRE:

# validate inlining processing
import Base: @constprop

@constprop :none @inline isType(@nospecialize(t)) = throw("invalid inlining processing detected")
@constprop :none @noinline isType(i::Integer) = (println(IOBuffer(), "prevent inlining"); false)
let
    invoke(xs) = isType(xs[1])
    @test invoke(Any[10]) === false
end

@constprop :aggressive @inline isType(c, @nospecialize(t)) = c && throw("invalid inlining processing detected")
@constprop :aggressive @noinline isType(c, i::Integer) = c && (println(IOBuffer(), "prevent inlining"); false)
let
    invoke(xs) = isType(true, xs[1])
    @test invoke(Any[10]) === false
end

[aviatesk]

@nanosoldier runbenchmarks("foldl" || "sort" || "union", vs = ":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here.

[aviatesk]

@nanosoldier runbenchmarks("foldl" || "sort" || "union", vs = ":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here.

[aviatesk]

This PR should be solid on my responsibility, and I'm going to merge (and then backport to v1.7) once I confirmed green CIs and no suspicious logs.