Allow duplicate metric names

[jaydeluca]

Resolves #696

The goal of this change is to be able to have multiple metrics with the same name but different labels, which had previously been prevented by a check during registration.

While implementing this, I operated under a few assumptions:

• We will allow metrics with different labels to have the same name, but registering metrics with the same name and same labels results in an exception
• Metrics with the same name but different metric types will result in an exception
• Metrics with the same name but different labels should all be merged under a single HELP/TYPE heading

I originally attempted to implement the validation during metric registration, but that required doing some things that felt wrong, so I moved the validation to scrape time. This does result in a bit of a performance hit (~10% reduction in op/s in the benchmarks I ran) on the scrape operation.

The benchmarks I ran were on an apple M4. I did 4 runs total - 2 of the main branch for a baseline and 2 for this feature branch and then averaged them.

Benchmark results

TextFormatUtilBenchmark

Benchmark	Main #1	Main #2	Feature #1	Feature #2	Main Avg	Feature Avg
openMetricsWriteToByteArray	1,179,213 ± 5,959	1,170,787 ± 5,579	1,080,037 ± 5,836	1,068,906 ± 5,288	1,175,000	1,074,471
openMetricsWriteToNull	1,197,768 ± 16,513	1,191,845 ± 13,897	1,083,832 ± 3,873	1,076,979 ± 4,816	1,194,806	1,080,406
prometheusWriteToByteArray	1,281,014 ± 21,437	1,273,552 ± 26,889	1,144,375 ± 3,303	1,138,252 ± 5,744	1,277,283	1,141,313
prometheusWriteToNull	1,254,177 ± 7,513	1,308,925 ± 14,326	1,167,007 ± 6,183	1,152,662 ± 2,565	1,281,551	1,159,835

---

CounterBenchmark

Benchmark	Main #1	Main #2	Feature #1	Feature #2	Main Avg	Feature Avg
codahaleIncNoLabels	71,040 ± 1,676	81,355 ± 9,287	87,105 ± 8,193	78,829 ± 12,370	76,197	82,967
openTelemetryAdd	1,472 ± 38	1,350 ± 39	1,419 ± 57	1,350 ± 39	1,411	1,384
openTelemetryInc	1,437 ± 45	1,361 ± 82	1,389 ± 50	1,349 ± 68	1,399	1,369
openTelemetryIncNoLabels	1,521 ± 42	1,347 ± 88	1,330 ± 103	1,419 ± 74	1,434	1,375
prometheusAdd	84,285 ± 102	84,233 ± 79	84,015 ± 79	84,255 ± 28	84,259	84,135
prometheusInc	85,739 ± 160	86,172 ± 225	85,484 ± 68	86,128 ± 113	85,956	85,806
prometheusNoLabelsInc	85,586 ± 242	86,075 ± 65	85,385 ± 52	85,954 ± 57	85,830	85,669
simpleclientAdd	8,730 ± 1,429	9,114 ± 1,362	8,003 ± 733	9,200 ± 1,202	8,922	8,602
simpleclientInc	7,655 ± 798	10,469 ± 1,635	9,201 ± 1,476	9,746 ± 1,125	9,062	9,473
simpleclientNoLabelsInc	8,328 ± 908	9,124 ± 847	7,951 ± 815	9,988 ± 765	8,726	8,970

---

HistogramBenchmark

Benchmark	Main #1	Main #2	Feature #1	Feature #2	Main Avg	Feature Avg
openTelemetryClassic	780 ± 66	932 ± 75	794 ± 12	841 ± 98	856	817
openTelemetryExponential	737 ± 65	764 ± 96	732 ± 68	725 ± 51	751	729
prometheusClassic	3,844 ± 63	3,886 ± 117	4,080 ± 137	3,865 ± 62	3,865	3,973
prometheusNative	2,829 ± 245	2,933 ± 73	2,961 ± 148	2,969 ± 98	2,881	2,965
simpleclient	15,192 ± 1,914	11,926 ± 1,209	14,669 ± 395	15,781 ± 1,365	13,559	15,225

[zeitlinger]

great 😄