Materialize v0.36

v0.36.0

Add mz_internal.mz_sink_status and mz_internal.mz_sink_status_history to the system catalog. These objects respectively expose the current and historical state for each sink in the system, including potential error messages and additional metadata helpful for debugging.
Add mz_internal.mz_cluster_replica_sizes to the system catalog. This table provides a mapping of logical sizes (e.g. xlarge) to the number of processes, as well as CPU and memory allocations for each process. To monitor the resource utilization for all extant cluster replicas as a % of the total allocation, you can now use:
```
SELECT
  r.id AS replica_id,
  m.process_id,
  m.cpu_nano_cores / s.cpu_nano_cores * 100 AS cpu_percent,
  m.memory_bytes / s.memory_bytes * 100 AS memory_percent
FROM mz_cluster_replicas AS r
JOIN mz_internal.mz_cluster_replica_sizes AS s ON r.size = s.size
JOIN mz_internal.mz_cluster_replica_metrics AS m ON m.replica_id = r.id;
```
It’s important to note that these tables are part of an unstable interface of Materialize (mz_internal), which means that their values may change at any time, and you should not rely on them for tasks like capacity planning for the time being.
Add mz_catalog.mz_aws_privatelink_connections to the system catalog. This table contains a row for each AWS PrivateLink connection in the system, and allows you to retrieve the AWS principal that Materialize will use to connect to the VPC endpoint.
Return an error rather than crashing if the value of the AVRO KEY FULLNAME or AVRO VALUE FULLNAME option in an Avro-formatted Kafka sink is not a valid Avro name (#16433).
Return the current timestamp of the EpochMillis timeline when the mz_now () function is used outside the context of a specific timeline, such as SELECT mz_now();. The old behavior was to return u64::MAX.

Patch releases

v0.36.2

Fix incorrect decoding of negative timestamps (i.e. prior to the Unix epoch: January 1st, 1970 at 00:00:00 UTC) in Avro records (#16609).