feat(firestore): Adding vector search

[bhshkh]

b/346827360

Python and Nodejs already have support for vector search. More details can be found here

This PR introduces the same functionality in Go client library.

Sample Usage:
Save documents with vector field:

type CoffeeBeans struct {
	EmbeddedField firestore.Vector64
}
.......
........
docRef1 := client.Collection(vectorCollection).Doc("Arabica")
docRef1.Create(ctx, CoffeeBeans{
EmbeddedField: firestore.Vector64([]float64{1.0, 2.0, 3.0}),
})

docRef2 := client.Collection(vectorCollection).Doc("Excelsa")
docRef2.Create(ctx, CoffeeBeans{
EmbeddedField: firestore.Vector64([]float64{4, 5, 6}),
})

docRef3 := client.Collection(vectorCollection).Doc("Liberica")
docRef3.Create(ctx, CoffeeBeans{
EmbeddedField: firestore.Vector64([]float64{400, 500, 600}),
})

FindNearest query:

func findNearest(client *firestore.Client, ctx context.Context) error {
	vectorQuery := client.Collection(vectorCollection).FindNearest("EmbeddedField", firestore.Vector64{1, 2, 3}, firestore.FindNearestOpts{
		Limit:   2,
		Measure: firestore.DistanceMeasureEuclidean,
	})

	iter := vectorQuery.Documents(ctx)
	gotDocs, _ := iter.GetAll()

	for _, doc := range gotDocs {
		bean := CoffeeBeans{}
		err := doc.DataTo(&bean)
		fmt.Printf("findNearest bean: %+v, err: %+v\n", bean, err)

                docMap := doc.Data()
	}
	return nil
}

The above query will return Excelsa and Arabica beans

Major changes:

1. Added Vector64 type which is just an alias for []float64 and similarly Vector32
2. Added VectorQuery type which is a wrapper for query but allows only the Documents method.
3. Renamed variables in setReflectFromProtoValue function in firestore/from_value.go
4. Added vector type handing in setReflectFromProtoValue method. This is used in the DataTo method demonstrated above where the user passes in a struct and the Firestore document is populated into the struct
5. Added vector type handing in createFromProtoValue method. This is used in the Data method demonstrated above which returns the Firestore document in the form of a map.

[jba]

Thanks for the extensive tests!

[jba]

Very close! Some minor tweaks.