Summary

This PR

• removes automatic UUID generation from the SupabaseVectorStore implementation (from_texts and add_texts methods)
• makes ID handling optional, bringing it in line with the TypeScript/JavaScript version

When IDs aren't provided, the database is now responsible for generating them instead of the library forcing UUID strings.

Why need this change

The current Python implementation has several limitations compared to the TypeScript version especially when I use SupabaseVectorStore.from_documents:

1. Inflexible ID types: Always generates UUID strings in from_texts method (which is called by from_documents method), preventing use of AUTO_INCREMENT, SERIAL, or custom ID strategies
2. Database design constraints: Forces tables to use UUID string primary keys instead of integers or other types
3. Inconsistent behavior: TypeScript and Python versions handle IDs differently, causing confusion for users switching between languages
4. Performance overhead: Generates UUIDs even when the database could handle ID generation more efficiently
5. Limited customization: No way to use custom ID generation logic or database-native features in from_texts() method

What's the solution

1. Remove automatic UUID generation from add_texts() and from_texts() methods
2. Make IDs optional in add_vectors() and _add_vectors() methods
3. Add validation to ensure ID count matches document count when IDs are provided
4. Conditional ID inclusion - only add ID field to row data when explicitly provided
5. Remove unused imports - clean up uuid import

When IDs are provided:

vector_store = SupabaseVectorStore.from_documents(
    docs,
    embeddings,
    client=supabase,
    table_name="documents",
    query_name="match_documents",
    ids=["ids1", "id2", ...]
)

# NOW: ID: "ids1", "id2", ...
# BEFORE: ID: uuids generated by this library because from_texts doesn't respect the ids args. (it always overwrites `ids` with `[str(uuid.uuid4()) for _ in texts]`)
# reference: https://github.com/langchain-ai/langchain-community/blob/main/libs/community/langchain_community/vectorstores/supabase.py#L152

When IDs are NOT provided:

vector_store = SupabaseVectorStore.from_documents(
    docs,
    embeddings,
    client=supabase,
    table_name="documents",
    query_name="match_documents",
)

# ID: depends on your DB settings
# BEFORE: ID: uuids generated by this library