Add prometheus exemplar example

[dashpole]

Add an example of using the prometheus exporter to produce prometheus exemplars.

This PR does the following:

• Add tracerprovider initialization with a logging exporter
• Add trace instrumentation, which creates a span each loop
• Use a counter and histogram metric, instead of a gauge for the example. Instrumentation is taken from the OTLP example.

If I curl the endpoint using OpenMetrics, I can see exemplars

curl -H 'Accept: application/openmetrics-text; version=1.0.0; charset=utf-8' http://localhost:19090/metrics

# TYPE super_timer histogram
# HELP super_timer
super_timer_count{otel_scope_name="io.opentelemetry.example.prometheus"} 219.0 1676324922.849
super_timer_sum{otel_scope_name="io.opentelemetry.example.prometheus"} 219088.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="0.0"} 0.0 1676324922.849 # {span_id="b68bdb5d2557fe43",trace_id="72f259bbe3a8e9c7bc2885d5ad97e206"} 0.0 1676324922.020
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="5.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="10.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="25.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="50.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="75.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="100.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="250.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="500.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="750.0"} 0.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="1000.0"} 157.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="2500.0"} 219.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="5000.0"} 219.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="7500.0"} 219.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="10000.0"} 219.0 1676324922.849
super_timer_bucket{otel_scope_name="io.opentelemetry.example.prometheus",le="+Inf"} 219.0 1676324922.849
# TYPE example_counter counter
# HELP example_counter
example_counter_total{otel_scope_name="io.opentelemetry.example.prometheus"} 220.0 1676324922.849

The associated trace for the exemplar:

INFO: 'exampleSpan' : 72f259bbe3a8e9c7bc2885d5ad97e206 b68bdb5d2557fe43 INTERNAL [tracer: io.opentelemetry.example.prometheus:] AttributesMap{data={good=true, exampleNumber=218}, capacity=128, totalAddedValues=2}

cc @jsuereth

[dashpole]

I fixed the example by moving the histogram.record call inside the try block. (thanks @jack-berg !).

Now, the only problem i'm facing is that the value of the exemplar seems to be recorded incorrectly. All iterations are about 1000 milliseconds long, which can be seen in the histogram. All observations fall into the le="1000.0" or le="2500.0" buckets. But as you can see, the exemplar appears to have a value of 0.0, and is placed on the le="0.0" bucket.

[dashpole]

Using the logging exporter, I can confirm that it isn't an issue with the prometheus exporter:

INFO: metric: ImmutableMetricData{resource=Resource{schemaUrl=null, attributes={service.name="PrometheusExporterExample", telemetry.sdk.language="java", telemetry.sdk.name="opentelemetry", telemetry.sdk.version="1.22.0"}}, instrumentationScopeInfo=InstrumentationScopeInfo{name=io.opentelemetry.example.prometheus, version=null, schemaUrl=null, attributes={}}, name=super.timer, description=, unit=ms, type=HISTOGRAM, data=ImmutableHistogramData{aggregationTemporality=CUMULATIVE, points=[ImmutableHistogramPointData{getStartEpochNanos=1676326373284000000, getEpochNanos=1676326433302000000, getAttributes={}, getSum=59053.0, getCount=59, hasMin=true, getMin=1000.0, hasMax=true, getMax=1026.0, getBoundaries=[0.0, 5.0, 10.0, 25.0, 50.0, 75.0, 100.0, 250.0, 500.0, 750.0, 1000.0, 2500.0, 5000.0, 7500.0, 10000.0], getCounts=[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 31, 28, 0, 0, 0, 0], getExemplars=[ImmutableDoubleExemplarData{filteredAttributes={}, epochNanos=1676326429485000000, spanContext=ImmutableSpanContext{traceId=ef29eebb7cde692aa527d9c114b384af, spanId=e0a4334cc4242518, traceFlags=01, traceState=ArrayBasedTraceState{entries=[]}, remote=false, valid=true}, value=0.0}, ImmutableDoubleExemplarData{filteredAttributes={}, epochNanos=1676326432488000000, spanContext=ImmutableSpanContext{traceId=9c16ce50f3788156dc0fb8424810b7bb, spanId=94e329af9852cc7c, traceFlags=01, traceState=ArrayBasedTraceState{entries=[]}, remote=false, valid=true}, value=0.0}]}]}}

The values of both are 0.0, even though only higher buckets have any observations.

[dashpole]

Even though exemplar recording isn't currently working, I think this example is now correct, and should work after that is fixed. I consider this ready for review.

[dashpole]

(sorry, tests are passing now)

[trask]

thx!