Optimize tag handling

[sslotnick-isp]

This optimizes how tags are handled to minimize map to slice conversions. One of the nice things about this library is that tags can be specified as a map instead of having to munge tag keys and values together as is necessary with datadog-go. However, the current version converts between map and list on every metric call which leads to a lot of extra allocations. The main reason for maintaining the map at all was to ensure duplicate tags got overwritten. However, I don't know of any use cases where this kind of override is currently being used. Secondly, I tested how DataDog would handle duplicate tag keys and found that it will report both:

Note: It shows the values comma separated but if I were to filter for just testapp:abc I would get both the yellow and purple lines.

The savings are significant in memory and CPU time:

benchmark                                       old ns/op     new ns/op     delta
Benchmark_0Tags_100Emits-8                      15389         17235         +12.00%
BenchmarkTags_5Tags_100Emits-8                  178638        24261         -86.42%
BenchmarkTags_5Tags_100Emits_WithInline-8       311105        90284         -70.98%
BenchmarkTags_10Tags_1000Emits-8                2652560       349187        -86.84%
BenchmarkTags_10Tags_1000Emits_WithInline-8     3836957       1051966       -72.58%
BenchmarkTags_15Tags_100Emits-8                 386667        47259         -87.78%
BenchmarkTags_15Tags_100Emits_WithInline-8      618854        121621        -80.35%

benchmark                                       old allocs     new allocs     delta
Benchmark_0Tags_100Emits-8                      2              1              -50.00%
BenchmarkTags_5Tags_100Emits-8                  1603           17             -98.94%
BenchmarkTags_5Tags_100Emits_WithInline-8       2403           717            -70.16%
BenchmarkTags_10Tags_1000Emits-8                31005          32             -99.90%
BenchmarkTags_10Tags_1000Emits_WithInline-8     39068          7033           -82.00%
BenchmarkTags_15Tags_100Emits-8                 4605           47             -98.98%
BenchmarkTags_15Tags_100Emits_WithInline-8      5408           747            -86.19%

benchmark                                       old bytes     new bytes     delta
Benchmark_0Tags_100Emits-8                      80            48            -40.00%
BenchmarkTags_5Tags_100Emits-8                  32369         368           -98.86%
BenchmarkTags_5Tags_100Emits_WithInline-8       107579        51905         -51.75%
BenchmarkTags_10Tags_1000Emits-8                641021        707           -99.89%
BenchmarkTags_10Tags_1000Emits_WithInline-8     1699559       596099        -64.93%
BenchmarkTags_15Tags_100Emits-8                 98174         1012          -98.97%
BenchmarkTags_15Tags_100Emits_WithInline-8      260658        68551         -73.70%

Note: This also adds back the dep files so that teams not on go mods will still be able to upgrade.

[codecov-commenter]

Codecov Report

Merging #14 into master will increase coverage by 0.08%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #14      +/-   ##
==========================================
+ Coverage   93.80%   93.89%   +0.08%     
==========================================
  Files           6        6              
  Lines         420      426       +6     
==========================================
+ Hits          394      400       +6     
  Misses         18       18              
  Partials        8        8              

Impacted Files	Coverage Δ
metrics/datadog.go	94.87% <100.00%> (-0.26%)	⬇️
metrics/logger.go	97.40% <100.00%> (ø)
metrics/recorder.go	97.18% <100.00%> (ø)
metrics/util.go	90.24% <100.00%> (+2.36%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update af57a97...c02370b. Read the comment docs.

[danielgtaylor-isp]

LGTM 👍