Sampling: Each exporter should have isolated control over its sampler decision and custom processing

[kalyanaj]

[This is based on @jmacd's PR feedback in https://github.com/open-telemetry/opentelemetry-dotnet/pull/4206]

What are you trying to achieve?
With the current sampler and processor API, any decisions taken at the end of this pipeline can affect all exporters (if multiple exporters are configured).

What did you expect to see?
Each exporter should have isolated control over its sampler decision and custom processing.

Additional context.
[This is based on @jmacd's PR feedback in https://github.com/open-telemetry/opentelemetry-dotnet/pull/4206]:

"I feel there is an OTel specification-level problem here, possibly a discussion topic. I think this calls for a "v2" Sampler and Processor API that would correspond with the Metrics SDK's Reader. That is a way of specifying that each exporter should have isolated control over its sampler decision and custom processing and particularly that the Trace SDK should manage composition of sampler policies. See also #2179."