I believe there is a change of API or a performance issue when using batches with 2.1.0.

Environment details

• OS type and version: Windows 10
• Python version: 3.8.7
• pip version: 20.3.3
• google-cloud-datastore version: 2.1.0
• google-api-core version: 1.24.1

Steps to reproduce

1. Use DatastoreClient::put_multi with 500 entities inside a batch

Code example

with client.batch():
    client.put_multi(chunk)

Performance

When using 1.15.0, my code can handle 9 batches of 500 entities in about 1-2 seconds while 2.1.0 takes literal minutes.
Disclaimer: I use the datastore emulator

EDIT: It can be reproduced in 2.0.0