Cratedb-toolkit

- Fixed connectivity for `jobstats collect`
- Refactored code and improved CLI interface of `ctk info` vs. `ctk cfr`
- Dependencies: Updated to `crate-2.0.0`, which uses `orjson` for JSON marshalling
- CFR: Job statistics and slow-query exploration per Marimo notebook. Thanks, WalBeh.

- Dependencies: Minimize dependencies of core installation,
defer `polars` to `cratedb-toolkit[io]`.
- Fixed `ctk cfr info record` about too large values of `ulimit_hard`
- Improved `ctk shell` to also talk to CrateDB standalone databases
- Added basic utility command `ctk tail`, for tailing a database
table, and optionally following the tail
- Table Loader: Added capability to load InfluxDB Line Protocol (ILP) files
- Query Collector: Now respects `CRATEDB_SQLALCHEMY_URL` environment variable

- MongoDB: Added Zyp transformations to the CDC subsystem,
making it more symmetric to the full-load procedure.
- Query Converter: Added very basic expression converter utility with
CLI interface
- DynamoDB: Added query expression converter for relocating object
references, to support query migrations after the breaking change
with the SQL DDL schema, by v0.0.27.

- IO: Improved `BulkProcessor` when running per-record operations by
also checking `rowcount` for handling `INSERT OK, 0 rows` responses
- MongoDB: Fixed BSON decoding of `{"$date": 1180690093000}` timestamps
by updating to commons-codec 0.0.21.
- Testcontainers: Don't always pull the OCI image before starting.
It is unfortunate in disconnected situations.

- MongoDB: Updated to pymongo 4.9
- DynamoDB: Change CrateDB data model to use (`pk`, `data`, `aux`) columns
Attention: This is a breaking change.

- MongoDB: Configure `MongoDBCrateDBConverter` after updating to commons-codec 0.0.18
- DynamoDB CDC: Fix `MODIFY` operation to also propagate deleted attributes