perf: improve has_tabs_or_newline performance

[anonrig]

This seems to slightly reduce the instruction count for url_aggregator, but please run local benchmarks to verify

Before:

• ada_url -> instructions/url=2.55753k
• ada_url_aggregator -> instructions/url=2.76077k

After:

• ada_url -> instructions/url=2.55742k
• ada_url_aggregator -> instructions/url=2.76044k

[lemire]

This is fine, but I bet you can do it with a lambda. Note that you are doing just the NEON routine, you could do the others as well.

I expect that this will not change performance. However, it can make the code more concise, which is good.

[CarlosEduR]

@anonrig good to go