baggage: Member.String encodes only when necessary

[pellared]

Fixes #4773

I also added Add UTF-8 test cases and some benchmarks.

Benchmark:

goos: linux
goarch: amd64
pkg: go.opentelemetry.io/otel/baggage
cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz
BenchmarkNew-16                   668466              1628 ns/op             832 B/op         16 allocs/op
BenchmarkNewMember-16           26077026                45.27 ns/op            0 B/op          0 allocs/op
BenchmarkParse-16                 954313              1116 ns/op             832 B/op          7 allocs/op
BenchmarkString-16                569619              1893 ns/op             960 B/op         18 allocs/op
BenchmarkValueEscape/nothing_to_escape-16               135047982                7.684 ns/op           0 B/op          0 allocs/op
BenchmarkValueEscape/requires_escaping-16               28184155                42.72 ns/op           16 B/op          1 allocs/op
BenchmarkValueEscape/long_value-16                       1492291               987.6 ns/op           576 B/op          2 allocs/op

Benchstat (in old.txt the benchmarks use url.PathEscape instead of valueEscape):

goos: linux
goarch: amd64
pkg: go.opentelemetry.io/otel/baggage
cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz
                                 │   old.txt    │               new.txt               │
                                 │    sec/op    │   sec/op     vs base                │
New-16                              1.221µ ± 5%   1.262µ ± 7%        ~ (p=0.055 n=10)
NewMember-16                        35.38n ± 5%   35.50n ± 3%        ~ (p=0.811 n=10)
Parse-16                            963.8n ± 4%   929.9n ± 3%   -3.52% (p=0.019 n=10)
String-16                           2.183µ ± 6%   1.632µ ± 4%  -25.22% (p=0.000 n=10)
ValueEscape/nothing_to_escape-16   13.465n ± 6%   6.957n ± 4%  -48.33% (p=0.000 n=10)
ValueEscape/requires_escaping-16    67.95n ± 5%   36.34n ± 6%  -46.52% (p=0.000 n=10)
ValueEscape/long_value-16          1596.0n ± 5%   739.2n ± 5%  -53.68% (p=0.000 n=10)
geomean                             279.3n        199.8n       -28.48%

                                 │    old.txt    │               new.txt                │
                                 │     B/op      │    B/op     vs base                  │
New-16                              832.0 ± 0%     832.0 ± 0%        ~ (p=1.000 n=10) ¹
NewMember-16                        0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
Parse-16                            832.0 ± 0%     832.0 ± 0%        ~ (p=1.000 n=10) ¹
String-16                          1024.0 ± 0%     960.0 ± 0%   -6.25% (p=0.000 n=10)
ValueEscape/nothing_to_escape-16    0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
ValueEscape/requires_escaping-16    16.00 ± 0%     16.00 ± 0%        ~ (p=1.000 n=10) ¹
ValueEscape/long_value-16           640.0 ± 0%     576.0 ± 0%  -10.00% (p=0.000 n=10)
geomean                                        ²                -2.40%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                                 │   old.txt    │               new.txt               │
                                 │  allocs/op   │ allocs/op   vs base                 │
New-16                             16.00 ± 0%     16.00 ± 0%       ~ (p=1.000 n=10) ¹
NewMember-16                       0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=10) ¹
Parse-16                           7.000 ± 0%     7.000 ± 0%       ~ (p=1.000 n=10) ¹
String-16                          18.00 ± 0%     18.00 ± 0%       ~ (p=1.000 n=10) ¹
ValueEscape/nothing_to_escape-16   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=10) ¹
ValueEscape/requires_escaping-16   1.000 ± 0%     1.000 ± 0%       ~ (p=1.000 n=10) ¹
ValueEscape/long_value-16          2.000 ± 0%     2.000 ± 0%       ~ (p=1.000 n=10) ¹
geomean                                       ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

The results of String and ValueEscape show improvement.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (d23c060) 82.2% compared to head (005ba1b) 82.2%.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #4775   +/-   ##
=====================================
  Coverage   82.2%   82.2%           
=====================================
  Files        226     226           
  Lines      18344   18383   +39     
=====================================
+ Hits       15087   15126   +39     
  Misses      2975    2975           
  Partials     282     282           

Files	Coverage Δ
baggage/baggage.go	100.0% <100.0%> (ø)

[dmathieu]

Should you run benchstat to compare the benchmarks between main and your branch?

[pellared]

Should you run benchstat to compare the benchmarks between main and your branch?

I can do that today in a few hours or tomorrow morning.
The result should be similar (maybe a little faster as there should be a few CPU cycles less). I intentionally almost copied the code from url.PathEscape. I wanted to add the benchmarks if someone would want to improve it in future.

[pellared]

@dmathieu Benchstat results added.

[yurishkuro]

I have no objection to this as an optimization of existing incorrect behavior - its not supposed to encode anything at all, only Propagator should be doing the encoding.

[MrAlias]

This was merged without two approvals from the Go Approvals team.

[pellared]

@MrAlias Sorry, I have missed that @yurishkuro is not an approver 😬