Availability: Adds non-blocking cache to partition key ranges

[j82w]

Pull Request Template

Description

Issues being fixed:

1. The current implementation is once a refresh occurs all other requests are blocked until the refresh completes. Even if they have 1000 partitions and only 1 is splitting all 999 partitions will be blocked until the refresh completes.
2. If the refresh fails the old cache value is lost. Then the SDK has to reload the entire which cause additional load to the service on the meta data requests.
3. Removed cancellation token being passed in. Most scenario it was already set to none. The cancellation should not be passed into cache because multiple other requests are likely to be waiting on that result. If the first request cancels it would negatively impact the rest of the requests. It's better to let the cache finish then let the request fail after.

Type of change

Please delete options that are not relevant.

• [] Bug fix (non-breaking change which fixes an issue)
• [] New feature (non-breaking change which adds functionality)
• [] Breaking change (fix or feature that would cause existing functionality to not work as expected)
• [] This change requires a documentation update

Closing issues

To automatically close an issue: closes #3060