MA.operate!(+, pol1, pol2): prevent captured variable perf issue (#145

)

The local function was spuriously recursive, making it not recursive
fixes the performance problem. That is, this removes some run time
dispatch and decreases the amount of allocation, but there's still
some run time dispatch left in other places, according to JET.jl.

Before:

```julia-repl
julia> import MutableArithmetics; const MA = MutableArithmetics;

julia> using DynamicPolynomials

julia> @PolyVar x y;

julia> p = 2x + y;

julia> q = x + 2y;

julia> @allocated MA.operate!(+, p, q)
34413936
```

After:

```julia-repl
julia> import MutableArithmetics; const MA = MutableArithmetics;

julia> using DynamicPolynomials

julia> @PolyVar x y;

julia> p = 2x + y;

julia> q = x + 2y;

julia> @allocated MA.operate!(+, p, q)
30835632
```

Both REPL runs was with nightly Julia v1.11.

src/operators.jl

@@ -154,8 +154,9 @@ function MA.operate!(
         push!(p.a, t[1])
         return push!(p.x.Z, t[2])
     end
-    compare_monomials(t::_NoVarTerm, j::Int) = _exponents_compare(q, j, t[2])
-    compare_monomials(i::Int, j::Int) = compare_monomials(get1(i), j)
+    compare_monomials_impl(t, j) = _exponents_compare(q, j, t[2])
+    compare_monomials(t::_NoVarTerm, j::Int) = compare_monomials_impl(t, j)
+    compare_monomials(i::Int, j::Int) = compare_monomials_impl(get1(i), j)
     combine(i::Int, j::Int) = p.a[i] = MA.operate!!(op, p.a[i], q.a[j])
     combine(t::_NoVarTerm, j::Int) = (MA.operate!!(op, t[1], q.a[j]), t[2])
     function resize(n)