Match DeferredRender in performance

[stephannv]

Previous code has performance penalty compared to DeferredRender. This changes will match Slotable performance with DeferredRender.

ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [arm64-darwin23]
                  user     system      total        real
Deferred      3.732848   0.003853   3.736701 (  3.746092)
Slotable Old  4.628438   0.005811   4.634249 (  4.640234)
Slotable New  3.685357   0.003766   3.689123 (  3.693959)


ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [arm64-darwin23]
Warming up --------------------------------------
            Deferred    26.779k i/100ms
        Slotable Old    21.636k i/100ms
        Slotable New    27.013k i/100ms
Calculating -------------------------------------
            Deferred    267.884k (± 0.6%) i/s -      1.366M in   5.098391s
        Slotable Old    216.193k (± 0.4%) i/s -      1.082M in   5.003961s
        Slotable New    270.082k (± 0.5%) i/s -      1.351M in   5.001001s