Update cloudwatch_exporter to 0.59.0

[github-actions]

Update cloudwatch_exporter to 0.59.0.

Important news and breaking changes

This release brings a bunch of breaking changes:

• Setting roundingPeriod for discovery jobs is deprecated, a warning will be logged at startup. This is being deprecated in favor of always using the metric period. The implementation for roundingPeriod can result in inconsistent Start and EndTime between batches. This negates its intent to ensure Start and EndTimes align with the metric period for CloudWatch best practices. This has the potential to produce data which will look inaccurate when compared against CloudWatch itself driving a lot of confusion. See [BUG] Jobs using length, delay, and period at the job level and metric level results in inconsistent start time and end time for GetMetricData queries prometheus-community/yet-another-cloudwatch-exporter#1290 for further context.
• Setting delay at the metric level is deprecated, a warning will be logged at startup. This delay configuration has existed for a long time but was never utilized. Deprecating it and eventually removing it was chosen to simplify the configuration. See [BUG] Jobs using length, delay, and period at the job level and metric level results in inconsistent start time and end time for GetMetricData queries prometheus-community/yet-another-cloudwatch-exporter#1290 (comment) for further context.
• For discovery jobs, the type field and the keys of exportedTagsOnMetrics must be the AWS namespace rather than the alias (the README contains an up-to-date list of namespaces). Aliases are not allowed anymore. An error will be thrown at startup in an invalid namespace or an alias is used.
• Some metric names have been changed to avoid duplicating the namespace. This includes:
  • aws_es_esreporting_failed_request_sys_err_count is aws_es_reporting_failed_request_sys_err_count
  • aws_es_esreporting_failed_request_user_err_count is aws_es_reporting_failed_request_user_err_count
  • aws_es_esreporting_request_count is aws_es_reporting_request_count
  • aws_es_esreporting_success_count is aws_es_reporting_success_count
  • aws_kafka_kafka_app_logs_disk_used is aws_kafka_app_logs_disk_used
  • aws_kafka_kafka_data_logs_disk_used is aws_kafka_data_logs_disk_used
  • aws_rds_rdsto_aurora_postgre_sqlreplica_lag is aws_rds_to_aurora_postgre_sqlreplica_lag
  • aws_glue_glue_.* is aws_glue_.*

These breaking changes will allow making the configuration easier to understand and less error prone, and also to build better documentation around supported services.

Bugfixes and features

Features:

• Add AWS/SecretsManager to the services list by @taraspos
• Support partner events buses by @HristoStoyanovYotpo
• discovery.exportedTagsOnMetrics: validate that keys match one of the job types defined by @cristiangreco

Refactoring:

• Update comment in factory.go by @andriikushch
• getmetricdata: move window calculator to processor by @kgeckhart
• promutil: clean up prom metric names that duplicate parts of the namespace by @tristanburgess
• promutil: rewrite sanitisation funcs for memory optimisation by @cristiangreco
• Do not allow using aliases as job types in discovery jobs by @cristiangreco

Dependencies

• Bump github.com/aws/aws-sdk-go from 1.51.16 to 1.51.21
• Bump github.com/aws/aws-sdk-go-v2 group
• Bump github.com/prometheus/common from 0.52.2 to 0.52.3

New contributors

• @taraspos made their first contribution in Add AWS/SecretsManager to the services list prometheus-community/yet-another-cloudwatch-exporter#1330
• @HristoStoyanovYotpo made their first contribution in Support partner events buses prometheus-community/yet-another-cloudwatch-exporter#1359

Full Changelog: prometheus-community/yet-another-cloudwatch-exporter@v0.58.0...v0.59.0