Should long-running operations (LROs) have a common poll interval configuration? #1426
Description
While working on Test Proxy playback and record migration it was found that tests that call into long-running operations (LROs) aren't seeing much benefit using playback instead of live testing as the polling interval is statically configured (not by MCP code but we leverage default code paths which use a default polling interval).
We should investigate whether LRO polling interval should be hoisted to the common configurations like retry configurations, subscription ID, and more. With this both our testing and customers can decide their polling interval if needed rather than relying on the default provided.
Or, at least, we should look into a way for tests to smuggle in a polling interval during playback testing so a test with 30 10 second polls doesn't take 300 seconds waiting on nothing and instead could reduce the overall runtime to a few seconds, given playback polling isn't waiting for a server change to complete.