A new tool to create a vector search index based on:

• The index name.
• The field that contains the embedding.
• The dimensions of the embedding.
• The quantisation mechanism.
• The similarity function.
• The model to use for embeddings.
• An additional list of fields to pre-filter.

It should:

1. Check that the index name is valid
2. The field that contains the embedding is valid by:
    1. Checking it does in fact exist
    2. The type is valid for the model/dimensions/quantisation
3. The dimensions are valid for the specified model
4. The quantisation mechanism is valid for the specific model
5. The similarity function is valid for the model/quantisation
6. The additional list of fields to pre-filter do exist
   1. But shouldn't reject the index if a field for prefilter has not been found.

After all checks have been done, it must create an index and mention the model that was used for the index, the dimensions and the fields to pre-filter. If a field was non-existing, it should be mentioned too. Vector Search indexes are slow to create, so we should also mention this.