fix: performance regression in_span

[robertlaurin]

https://github.com/fastruby/fast-ruby/blob/main/code/proc-and-block/proc-call-vs-yield.rb

ruby proc-call-vs-yield.rb
Warming up --------------------------------------
          block.call     1.141M i/100ms
       block + yield     1.202M i/100ms
        unused block     1.495M i/100ms
               yield     1.389M i/100ms
Calculating -------------------------------------
          block.call     11.485M (± 1.4%) i/s -     58.175M in   5.066280s
       block + yield     11.910M (± 1.6%) i/s -     60.089M in   5.046587s
        unused block     14.945M (± 5.4%) i/s -     74.725M in   5.019902s
               yield     14.068M (± 1.0%) i/s -     70.858M in   5.037486s

Comparison:
        unused block: 14945101.0 i/s
               yield: 14067574.6 i/s - same-ish: difference falls within error
       block + yield: 11910103.8 i/s - 1.25x  slower
          block.call: 11485071.3 i/s - 1.30x  slower

Edit: to add this was with ruby 3.2.2 (2023-03-30 revision e51014f9c0) [arm64-darwin22]

We'll probably want to update the linters here https://github.com/open-telemetry/opentelemetry-ruby/pull/1323/files#diff-17159f63bf877cf73b9013894bea9311fb40cd7a4af890df4b0c77ba9fe2828b

[arielvalentin]

Anyone working on core have any insights into what the cause of the latency is in these execution paths?

Are the Rubocop maintainers aware?