Webrequest time-series pipeline

• pipeline to aggregate cost metrics of webrequests traffic into a time-series dataset, with configurable grouping columns and window size
• method to create lancedb tables with vector search indices, backed by ceph s3 storage
• transformation to run similarity search using lacedb distributed by spark
• transformation to use push-down filters for read datasets partitioned year/month/day/hour
• update readme for airflow dev on k8s
• unit tests