Local mode: wrong-kind query vector reports the named vector as 'not found' (e.g. 1-D query against a multivector) #1444

What happened?

In local mode, querying a collection with a vector of the wrong kind for the named vector produces a misleading ValueError.

When a collection has a named multivector and the query is a 1-D vector (or vice versa: a 2-D query against a dense named vector), the validation falls into the wrong branch and raises:

Dense vector mv is not found in the collection        # 'mv' is a multivector, 1-D query given
Multivector  is not found in the collection           # unnamed dense collection, 2-D query given

The vector does exist — the query format is wrong. The message even shows an empty name in the unnamed case, which makes debugging worse.

Reproduction (local mode, qdrant-client main)

from qdrant_client import QdrantClient, models

client = QdrantClient(path=":memory:")
client.create_collection("m", vectors_config={
    "mv": models.VectorParams(size=2, distance=models.Distance.COSINE,
        multivector_config=models.MultiVectorConfig(comparator=models.MultiVectorComparator.MAX_SIM))})
client.upsert("m", points=[models.PointStruct(id=1, vector={"mv": [[1.0, 0.0]]})])

client.query_points("m", query=[1.0, 0.0], using="mv", limit=5)
# ValueError: Dense vector mv is not found in the collection  <- misleading

Server-side correctly requires a 2-D query for multivector collections (per the docs), so the local rejection itself is right — the error is just wrong about the reason.

Suggested fix

When the named vector exists but as a different kind, raise an error that says so:

Vector 'mv' is a multivector vector, but the query vector is a single vector (1-D).
A multivector collection requires a matching query format.

Truly missing names keep the original "not found" message. Happy to open a PR with the fix + tests.