Health Imaging StartDICOMImportJob always processes entire bucket regardless of the input uri.

[LeoReubelt]

I have an application that uses Health Imaging.
And I have a lambda that starts import jobs with the following function.

         const command = new StartDICOMImportJobCommand({
            clientToken: uuid.v4(),
            inputOwnerAccountId: "",
            datastoreId: dataStoreId,
            dataAccessRoleArn:
                'arn:aws:iam::' +
                process.env.ACCOUNT +
                ':role/ADD_DICOM_TO_DATASTORE_ROLE-' +
                process.env.ENV,
            inputS3Uri: inputPath,
            outputS3Uri: outputPath,
            jobName: awsRequestId,
         });
        
        const response = await client.send(command);

I enter a specific folder for inputS3Uri.
And it has been working great.
But I started getting the following error.

DICOM import job has failed.
FILE_COUNTS_LIMIT_EXCEED: Input bucket contains more than 5000 files.

And it seems like it is processing the entire bucket. Not just the folder that I pass in.
If I go to the Data Store import job in the console and click on the link for the 'Import source location in S3', i get the following error...

The bucket named "xxxxxx" can't be found in the US East (Ohio) us-east-2 Region. The bucket named "xxxxx" is in the US East (N. Virginia) us-east-1 Region.

And then it redirects to the correct bucket, but not to the folder specified and not to the root, but to a folder in between.
I have confirmed that the path is correct and that the files at the path exist.
Also everything in our application is us-east-1. There are no references to us-east-2. And I am explicitly specifying us-east-1 when the clients are created.

const s3 = new AWS.S3({
    region: process.env.REGION,
});
const ahi = new AWS.MedicalImaging({
    region: process.env.REGION,
});

So, why is health imaging processing the entire bucket, and where is it getting us-east-2 from?