fix: cancel retries on shutdown

[pichlermarc]

Which problem is this PR solving?

With the fix from #6147, we started re-trying failed exports more aggressively. The result is that shutdown now hangs for a couple of seconds before it finishes. (ref: open-telemetry/opentelemetry-js-contrib#3349)

This PR introduces several changes to address this:

• OTLP exporters now cancel in-flight retries when exporter.forceFlush() is called
• BatchSpanProcessor#shutdown() now schedules exports, but calls SpanExporter#forceFlush() right after scheduling the export, signaling the exporter to speed up
• BatchLogRecordProcessor#shutdown() does the same
• PeriodicExportingMetricReader#shutdown() does the same

The diff for this is fairly large, and I don't expect anybody to review this PR, so I'll break this down into smaller chunks, keeping this PR in draft to show all changes working together.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

• Existing tests (had to make some changes as the previous tests were relying heavily on behvaior that was not guaranteed by the API)
• Added test in @opentelemetry/sdk-node to test the same shutdown code path used in @opentelemetry/auto-instrumentation-node

[pichlermarc]

I'm spinning off mutiple PRs from this one, the first one is #6356