Better exception handling

[fhibf]

Description

Adding better exception handling for a few situations:

• Extending PreconditionFailedException to support inner exceptions.
• Raising PreconditionFailedException when an operation fails due a partition reaching 20GB. That should replace HTTP500s with HTTP400s, indicating to customer that a failure on their end caused the issue.
• Adding hew FHIR Exception type called TransactionFailureException to replace odd InvalidOperationExceptions raised in case of Cosmos DB transaction failures.
• Adding more logs for when FHIR profiles are refreshed by BundleHandler. Additionally, logging for how long the refresh ran and handling cases when the refresh fails. Fails during the profile refresh should not affect the response of bundles.
• Handling null reference exceptions at Bundle SanitizeString. I noticed those errors recently, and I want to avoid the noise caused by these errors. These errors are not currently affecting the execution of bundles.

Related issues

AB#134896.

Testing

Describe how this change was tested.

Semver Change (docs)

Patch|Skip|Feature|Breaking (reason)