baggage: more efficient member validation

[kevinburkesegment]

Instead of performing many, many comparisons for each character in a key and value, create an array of 128 bits and test for membership in that array, which is quicker.

Benchmarks on a Mac M1, ran 20 times.

name                             old time/op    new time/op    delta
New-8                               747ns ± 4%     741ns ± 3%    ~     (p=0.089 n=17+19)
NewMemberRaw-8                     11.4ns ± 1%    10.9ns ± 1%  -4.67%  (p=0.000 n=17+19)
Parse-8                             619ns ± 2%     603ns ± 1%  -2.54%  (p=0.000 n=18+18)
String-8                           1.12µs ± 4%    1.09µs ± 2%  -2.65%  (p=0.000 n=19+18)
ValueEscape/nothing_to_escape-8    6.93ns ± 1%    6.94ns ± 0%  +0.18%  (p=0.012 n=18+20)
ValueEscape/requires_escaping-8    26.3ns ± 2%    26.1ns ± 1%  -0.91%  (p=0.000 n=19+19)
ValueEscape/long_value-8            641ns ± 4%     595ns ± 1%  -7.14%  (p=0.000 n=18+17)

The approach is taken from the same approach in encoding/json in the standard library, which I also contributed.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.5%. Comparing base (ed666f7) to head (2498094).

Additional details and impacted files

@@           Coverage Diff           @@
##            main   #5214     +/-   ##
=======================================
- Coverage   84.5%   84.5%   -0.1%     
=======================================
  Files        258     258             
  Lines      17110   17096     -14     
=======================================
- Hits       14468   14452     -16     
- Misses      2343    2345      +2     
  Partials     299     299             

Files	Coverage Δ
baggage/baggage.go	98.1% <100.0%> (-0.1%)	⬇️

... and 1 file with indirect coverage changes

[kevinburkesegment]

I don't think this PR needs a changelog entry.

[XSAM]

I am ok with this

[pellared]

I don't think this PR needs a changelog entry.

We tend to add changelog entries when we improve performance.

[pellared]

The approach is taken from the same approach in encoding/json in the standard library, which I also contributed.

Can you please provide a reference? I cannot find anything here: https://github.com/golang/go/pulls/kevinburkesegment

[kevinburkesegment]

Can you please provide a reference? I cannot find anything here: https://github.com/golang/go/pulls/kevinburkesegment

Ah, this is my work account - all of my Go contributions are on the email under my other Github username, kevinburke, but also Google does not use Github for pull requests.

That commit is here: golang/go@ed8f207

[pellared]

@open-telemetry/go-approvers waiting 1 day before merging.