Although you appear to have made (thank you!) a diligent effort to present data to explain your point, I’m afraid that I see the various alternatives to be “apples and oranges.” As I superficially review your four alternatives – and without expanding your post – I see what appears to be a “side-by-side comparison” of four very different algorithms. Also, you neglect to place the three “Perl alternatives” (chomp, chop, neither ...) into the benchmark. Could you please do that, and post the results as an additional comment to this thread? Please show us 4 + 2 = 6 lines of benchmark-data.
(And if I “just missed it,” I apologize in advance.)