test: unify json & parquet checkpoint creation methods in the MockTable

[sebastiantia]

Please describe why this is necessary.

Our current test suite uses two distinct methods to create checkpoints in MockTable:
The method parquet_checkpoint accepts engine data
The method json_checkpoint accepta actions to create a JSON checkpoint file.

This inconsistency causes the test cases—such as in test_create_checkpoint_stream_reads_parquet_checkpoint_batch versus test_create_checkpoint_stream_reads_json_checkpoint_batch — to look very different, which increases maintenance overhead and introduces potential discrepancies in handling different checkpoint formats.

Describe the functionality you are proposing.

Create a single method (e.g. create_checkpoint) for the MockTable struct test utility that accepts checkpoint actions as engine data. This method should internally handle the conversion for both JSON and Parquet formats, dependent on a passed parameter.

Instead of having one function that takes a batch of engine data and another that takes actions, standardize on one as the input. This change will streamline the test setup and ensure consistency across both checkpoint types.

Update tests accordingly

Additional context

No response

[sebastiantia]

We are moving away from adding the checkpointing functionality in the mock table until a broader conversation is held as to whether or not we want to continue with the MockTable testing infrastructure