Investigate and improve/fix handling of large S3 buckets

[samuelhwilliams]

While converting S3FileAdmin from boto to boto3 in #2552, I noticed that it probably performs very poorly for large S3 buckets due to collecting all files from the bucket at once for most operations.

We should think about how to handle buckets with millions+ objects, which may include changes to:

• UI
• pagination of S3 calls
• caching
• config to limit the 'maximum' number of objects Flask-Admin will try to retrieve.