Shift application to a data lake architecture

[jayhill]

We don't yet have clear use cases around which to build APIs, design our DynamoDB table and so on.

Instead, we're going to store the data as Iceberg/Parquet files in S3 and use Athena for queries. We'll drop daily files throughout the month and then run a compaction job (OPTIMIZE) to put everything into a monthly file at EoM.

Partitioning will be by facility ('buncombe', 'wake', etc.) and year+month.

Rework Terraform and Serverless Framework IaC definitions to accommodate this architecture.