Make ID column unique for input data classes

[frayle-ons]

The input and output dataclasses define required columns that need to be in the dataframes that are passed in and out of each vectorstore method/api.

The id column is essentially the unique id tracker for many of the input dataframes: embed, search and reverse_search. Currently there are no limitations to the actual column content. If the user has non-unique values in this column it causes issues with the server rest_api processing. Also it's intuitive and clear to have unique ID values for each row. The pandera models should be updated to enforce this, or another appropriate way

[lukeroantreeONS]

If we're updating these input objects anyway, it might be nice to make it optional for the user to specify the id value(s) themselves instead of required - and have it default to [1,2,3,...] if not specified.

For the uniqueness validation, this can be done within the panderas schema.
E.g. for the Search Input:

_schema: pa.DataFrameSchema = pa.DataFrameSchema(
    {
        "id": pa.Column(str),
        "query": pa.Column(str),
    },
    coerce=True,
    checks=pa.Check.unique_columns(["id"])
)