[Instrumentation.Runtime] Add a metric for total paused duration in GC for .NET 7 and greater versions

[xiang17]

Fixes #1163.

Changes

Add a new metric for total paused duration in GC process.runtime.dotnet.gc.duration with data from GC.GetTotalPauseDuration Method.

It is recommended over the time-in-gc EventCounter in the Well-known EventCounters in .NET.

The metrics would show in the output of Prometheus Exporter HttpListener (for how to set it up, follow this getting-started guide) like this:

# TYPE process_runtime_dotnet_gc_duration_ns counter
# UNIT process_runtime_dotnet_gc_duration_ns ns
# HELP process_runtime_dotnet_gc_duration_ns The total amount of time paused in GC since the process start.
process_runtime_dotnet_gc_duration_ns 8860000 1686806670504

For significant contributions please make sure you have completed the following items:

• Appropriate CHANGELOG.md updated for non-trivial changes
• Design discussion issue Runtime instrumentation: Can time-in-gc metric be published ? #1163
• Changes in public API reviewed: N/A

[codecov]

Codecov Report

Merging #1239 (3821fba) into main (dbfeaea) will decrease coverage by 0.62%.
The diff coverage is 100.00%.

❗ Current head 3821fba differs from pull request most recent head 2cedbec. Consider uploading reports for the commit 2cedbec to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1239      +/-   ##
==========================================
- Coverage   73.61%   72.99%   -0.62%     
==========================================
  Files         265      265              
  Lines        9346     9287      -59     
==========================================
- Hits         6880     6779     -101     
- Misses       2466     2508      +42     

Impacted Files	Coverage Δ
...ry.Exporter.Geneva/Metrics/GenevaMetricExporter.cs	70.03% <ø> (-4.52%)	⬇️
...elemetry.Instrumentation.Runtime/RuntimeMetrics.cs	100.00% <100.00%> (ø)

... and 1 file with indirect coverage changes