STS errors when many aws_s3 sinks are used

[Commerciallic]

Hi,
We're using vector 0.45.0 on an AWS machine.
We have many config files (several hundreds) that use an aws_s3 sink, using assume_role for uploading data to customers' S3 buckets.
When we increase the number of config files over ~200 (all sinks use the same assume_role), we start seeing STS timeouts:

STS returned an error assuming web identity role
dispatch failure: timeout: minimum throughput was specified at 1 B/s, but throughput of 0 B/s was observed (DispatchFailure(DispatchFailure { source: ConnectorError { kind: Timeout, source: ThroughputBelowMinimum { expected: Throughput { bytes_read: 1, per_time_elapsed: 1s }, actual: Throughput { bytes_read: 0, per_time_elapsed: 1s } }, connection: Unknown } }))

Does someone know why we should get timeouts when many assume_role requests are sent?
Are there perhaps relevant rate limits?

[Commerciallic]

This error happens because of stalled_stream_protection_config being enabled somewhere.

To fix, I downloaded vector's code and added .stalled_stream_protection(StalledStreamProtectionConfig::disabled()) where SdkConfig::builder() is used.
This did help, and the number of config files I could have without the throughput error raised, but didn't dissappear.
Looking at vector's dependencies, there are other places where SdkConfig::builder() is used but I have no control over it from vector's code.

Does someone have an idea how I can globally disable this annoying "feature"?
Any help would be greatly appreciated!

cc @jszwedko