Python: Fix Cosmos DB for MongoDB vector index kind

[EsraaKamel11]

Motivation and Context

CosmosMongoCollection._get_index_definitions populated cosmosSearchOptions["kind"] from the distance-function map
(a similarity code like "COS") instead of the index-kind map. As a result the created vector index used an invalid
kind, kind equalled similarity, and all HNSW/IVF/DiskANN tuning options (m, efConstruction, numList,
maxDegree, lBuild) were silently dropped because the match index_kind block could never match a vector-* case.

Cosmos DB for MongoDB vCore requires cosmosSearchOptions.kind ∈ {vector-ivf, vector-hnsw, vector-diskann}, so index
creation also fails against a live account.

Fixes #14104

Description

One-line fix: use INDEX_KIND_MAP_MONGODB[field.index_kind] for the index kind, keeping the distance-function map
for similarity (mirrors the NoSQL path, which already does this correctly). INDEX_KIND_MAP_MONGODB's value was
previously unused.

• Before: kind="COS", similarity="COS", tuning options dropped
• After: kind="vector-hnsw", similarity="COS", tuning options forwarded

Tests: updated the existing index test (which asserted the buggy "COS" value) and added a test verifying HNSW tuning
options (m, efConstruction) reach cosmosSearchOptions.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the SK Contribution Guidelines and the pre-submission formatting script raises no violations
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone 😄

[Copilot]

Pull request overview

Fixes Azure Cosmos DB for MongoDB vector index creation by correctly populating cosmosSearchOptions["kind"] from the index-kind mapping (e.g., vector-hnsw) while keeping cosmosSearchOptions["similarity"] sourced from the distance-function mapping (e.g., COS). This aligns the MongoDB vCore path with the existing NoSQL path and ensures vector index tuning options are applied.

Changes:

• Corrected _get_index_definitions to use INDEX_KIND_MAP_MONGODB[field.index_kind] for cosmosSearchOptions["kind"].
• Updated the existing unit test to assert the correct kind/similarity values.
• Added a unit test verifying HNSW tuning options (m, efConstruction) are forwarded into cosmosSearchOptions.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
python/semantic_kernel/connectors/azure_cosmos_db.py	Fixes MongoDB vector index kind mapping so index creation uses valid vector-* kinds and tuning options apply.
python/tests/unit/connectors/memory/azure_cosmos_db/test_azure_cosmos_db_mongodb_collection.py	Updates/extends tests to validate correct kind vs similarity and verify HNSW tuning options propagation.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[EsraaKamel11]

@microsoft-github-policy-service agree